RGBA to and from I420 with C implementation.

BUG=78
TEST=planar_test
Review URL: https://webrtc-codereview.appspot.com/798007

git-svn-id: http://libyuv.googlecode.com/svn/trunk@355 16f28f9a-4ce2-e073-06de-1de4eb20be90

README.chromium

@@ -1,6 +1,6 @@
 Name: libyuv
 URL: http://code.google.com/p/libyuv/
-Version: 354
+Version: 355
 License: BSD
 License File: LICENSE
 

include/libyuv/convert.h

@@ -133,7 +133,7 @@ int ABGRToI420(const uint8* src_frame, int src_stride_frame,
                uint8* dst_v, int dst_stride_v,
                int width, int height);
 
-// RGBA little endian (rgba in memory) to I420.
+// RGBA little endian (abgr in memory) to I420.
 int RGBAToI420(const uint8* src_frame, int src_stride_frame,
                uint8* dst_y, int dst_stride_y,
                uint8* dst_u, int dst_stride_u,

include/libyuv/version.h

@@ -11,6 +11,6 @@
 #ifndef INCLUDE_LIBYUV_VERSION_H_  // NOLINT
 #define INCLUDE_LIBYUV_VERSION_H_
 
-#define LIBYUV_VERSION 354
+#define LIBYUV_VERSION 355
 
 #endif  // INCLUDE_LIBYUV_VERSION_H_  NOLINT

source/convert.cc

@@ -1095,6 +1095,63 @@ int ABGRToI420(const uint8* src_abgr, int src_stride_abgr,
   return 0;
 }
 
+int RGBAToI420(const uint8* src_rgba, int src_stride_rgba,
+               uint8* dst_y, int dst_stride_y,
+               uint8* dst_u, int dst_stride_u,
+               uint8* dst_v, int dst_stride_v,
+               int width, int height) {
+  if (!src_rgba ||
+      !dst_y || !dst_u || !dst_v ||
+      width <= 0 || height == 0) {
+    return -1;
+  }
+  // Negative height means invert the image.
+  if (height < 0) {
+    height = -height;
+    src_rgba = src_rgba + (height - 1) * src_stride_rgba;
+    src_stride_rgba = -src_stride_rgba;
+  }
+  void (*RGBAToYRow)(const uint8* src_rgba, uint8* dst_y, int pix);
+  void (*RGBAToUVRow)(const uint8* src_rgba0, int src_stride_rgba,
+                      uint8* dst_u, uint8* dst_v, int width);
+
+  RGBAToYRow = RGBAToYRow_C;
+  RGBAToUVRow = RGBAToUVRow_C;
+#if defined(HAS_RGBATOYROW_SSSE3)
+  if (TestCpuFlag(kCpuHasSSSE3)) {
+    if (width > 16) {
+      RGBAToUVRow = RGBAToUVRow_Any_SSSE3;
+      RGBAToYRow = RGBAToYRow_Any_SSSE3;
+    }
+    if (IS_ALIGNED(width, 16)) {
+      RGBAToUVRow = RGBAToUVRow_Unaligned_SSSE3;
+      RGBAToYRow = RGBAToYRow_Unaligned_SSSE3;
+      if (IS_ALIGNED(src_rgba, 16) && IS_ALIGNED(src_stride_rgba, 16)) {
+        RGBAToUVRow = RGBAToUVRow_SSSE3;
+        if (IS_ALIGNED(dst_y, 16) && IS_ALIGNED(dst_stride_y, 16)) {
+          RGBAToYRow = RGBAToYRow_SSSE3;
+        }
+      }
+    }
+  }
+#endif
+
+  for (int y = 0; y < height - 1; y += 2) {
+    RGBAToUVRow(src_rgba, src_stride_rgba, dst_u, dst_v, width);
+    RGBAToYRow(src_rgba, dst_y, width);
+    RGBAToYRow(src_rgba + src_stride_rgba, dst_y + dst_stride_y, width);
+    src_rgba += src_stride_rgba * 2;
+    dst_y += dst_stride_y * 2;
+    dst_u += dst_stride_u;
+    dst_v += dst_stride_v;
+  }
+  if (height & 1) {
+    RGBAToUVRow(src_rgba, 0, dst_u, dst_v, width);
+    RGBAToYRow(src_rgba, dst_y, width);
+  }
+  return 0;
+}
+
 int RGB24ToI420(const uint8* src_rgb24, int src_stride_rgb24,
                 uint8* dst_y, int dst_stride_y,
                 uint8* dst_u, int dst_stride_u,
@@ -1763,6 +1820,14 @@ int ConvertToI420(const uint8* sample, size_t sample_size,
                      v, v_stride,
                      dst_width, inv_dst_height);
       break;
+    case FOURCC_RGBA:
+      src = sample + (src_width * crop_y + crop_x) * 4;
+      r = RGBAToI420(src, src_width * 4,
+                     y, y_stride,
+                     u, u_stride,
+                     v, v_stride,
+                     dst_width, inv_dst_height);
+      break;
     case FOURCC_RGBP:
       src = sample + (src_width * crop_y + crop_x) * 2;
       r = RGB565ToI420(src, src_width * 2,

source/convert_argb.cc

@@ -252,38 +252,6 @@ int I400ToARGB(const uint8* src_y, int src_stride_y,
   return 0;
 }
 
-int ABGRToARGB(const uint8* src_abgr, int src_stride_abgr,
-               uint8* dst_argb, int dst_stride_argb,
-               int width, int height) {
-  if (!src_abgr || !dst_argb ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    src_abgr = src_abgr + (height - 1) * src_stride_abgr;
-    src_stride_abgr = -src_stride_abgr;
-  }
-  void (*ABGRToARGBRow)(const uint8* src_abgr, uint8* dst_argb, int pix) =
-      ABGRToARGBRow_C;
-#if defined(HAS_ABGRTOARGBROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) &&
-      IS_ALIGNED(width, 4) &&
-      IS_ALIGNED(src_abgr, 16) && IS_ALIGNED(src_stride_abgr, 16) &&
-      IS_ALIGNED(dst_argb, 16) && IS_ALIGNED(dst_stride_argb, 16)) {
-    ABGRToARGBRow = ABGRToARGBRow_SSSE3;
-  }
-#endif
-
-  for (int y = 0; y < height; ++y) {
-    ABGRToARGBRow(src_abgr, dst_argb, width);
-    src_abgr += src_stride_abgr;
-    dst_argb += dst_stride_argb;
-  }
-  return 0;
-}
-
 // Convert BGRA to ARGB.
 int BGRAToARGB(const uint8* src_bgra, int src_stride_bgra,
                uint8* dst_argb, int dst_stride_argb,
@@ -317,6 +285,39 @@ int BGRAToARGB(const uint8* src_bgra, int src_stride_bgra,
   return 0;
 }
 
+// Convert ABGR to ARGB.
+int ABGRToARGB(const uint8* src_abgr, int src_stride_abgr,
+               uint8* dst_argb, int dst_stride_argb,
+               int width, int height) {
+  if (!src_abgr || !dst_argb ||
+      width <= 0 || height == 0) {
+    return -1;
+  }
+  // Negative height means invert the image.
+  if (height < 0) {
+    height = -height;
+    src_abgr = src_abgr + (height - 1) * src_stride_abgr;
+    src_stride_abgr = -src_stride_abgr;
+  }
+  void (*ABGRToARGBRow)(const uint8* src_abgr, uint8* dst_argb, int pix) =
+      ABGRToARGBRow_C;
+#if defined(HAS_ABGRTOARGBROW_SSSE3)
+  if (TestCpuFlag(kCpuHasSSSE3) &&
+      IS_ALIGNED(width, 4) &&
+      IS_ALIGNED(src_abgr, 16) && IS_ALIGNED(src_stride_abgr, 16) &&
+      IS_ALIGNED(dst_argb, 16) && IS_ALIGNED(dst_stride_argb, 16)) {
+    ABGRToARGBRow = ABGRToARGBRow_SSSE3;
+  }
+#endif
+
+  for (int y = 0; y < height; ++y) {
+    ABGRToARGBRow(src_abgr, dst_argb, width);
+    src_abgr += src_stride_abgr;
+    dst_argb += dst_stride_argb;
+  }
+  return 0;
+}
+
 // Convert RGBA to ARGB.
 int RGBAToARGB(const uint8* src_rgba, int src_stride_rgba,
                uint8* dst_argb, int dst_stride_argb,

source/convert_from.cc

@@ -828,6 +828,59 @@ int I420ToABGR(const uint8* src_y, int src_stride_y,
   return 0;
 }
 
+// Convert I420 to RGBA.
+int I420ToRGBA(const uint8* src_y, int src_stride_y,
+               const uint8* src_u, int src_stride_u,
+               const uint8* src_v, int src_stride_v,
+               uint8* dst_rgba, int dst_stride_rgba,
+               int width, int height) {
+  if (!src_y || !src_u || !src_v ||
+      !dst_rgba ||
+      width <= 0 || height == 0) {
+    return -1;
+  }
+  // Negative height means invert the image.
+  if (height < 0) {
+    height = -height;
+    dst_rgba = dst_rgba + (height - 1) * dst_stride_rgba;
+    dst_stride_rgba = -dst_stride_rgba;
+  }
+  void (*I422ToRGBARow)(const uint8* y_buf,
+                        const uint8* u_buf,
+                        const uint8* v_buf,
+                        uint8* rgb_buf,
+                        int width) = I422ToRGBARow_C;
+#if defined(HAS_I422TORGBAROW_NEON)
+  if (TestCpuFlag(kCpuHasNEON)) {
+    I422ToRGBARow = I422ToRGBARow_Any_NEON;
+    if (IS_ALIGNED(width, 16)) {
+      I422ToRGBARow = I422ToRGBARow_NEON;
+    }
+  }
+#elif defined(HAS_I422TORGBAROW_SSSE3)
+  if (TestCpuFlag(kCpuHasSSSE3) && width >= 8) {
+    I422ToRGBARow = I422ToRGBARow_Any_SSSE3;
+    if (IS_ALIGNED(width, 8)) {
+      I422ToRGBARow = I422ToRGBARow_Unaligned_SSSE3;
+      if (IS_ALIGNED(dst_rgba, 16) && IS_ALIGNED(dst_stride_rgba, 16)) {
+        I422ToRGBARow = I422ToRGBARow_SSSE3;
+      }
+    }
+  }
+#endif
+
+  for (int y = 0; y < height; ++y) {
+    I422ToRGBARow(src_y, src_u, src_v, dst_rgba, width);
+    dst_rgba += dst_stride_rgba;
+    src_y += src_stride_y;
+    if (y & 1) {
+      src_u += src_stride_u;
+      src_v += src_stride_v;
+    }
+  }
+  return 0;
+}
+
 // Convert I420 to RGB24.
 int I420ToRGB24(const uint8* src_y, int src_stride_y,
                 const uint8* src_u, int src_stride_u,
@@ -1228,6 +1281,14 @@ int ConvertFromI420(const uint8* y, int y_stride,
                      dst_sample_stride ? dst_sample_stride : width * 4,
                      width, height);
       break;
+    case FOURCC_RGBA:
+      r = I420ToRGBA(y, y_stride,
+                     u, u_stride,
+                     v, v_stride,
+                     dst_sample,
+                     dst_sample_stride ? dst_sample_stride : width * 4,
+                     width, height);
+      break;
     case FOURCC_BGGR:
       r = I420ToBayerBGGR(y, y_stride,
                           u, u_stride,

source/row.h

@@ -37,7 +37,9 @@ extern "C" {
 #define HAS_ABGRTOARGBROW_SSSE3
 #define HAS_ABGRTOUVROW_SSSE3
 #define HAS_ABGRTOYROW_SSSE3
-#define HAS_RGBATOARGBROW_SSSE3
+// TODO: #define HAS_RGBATOARGBROW_SSSE3
+// TODO: #define HAS_RGBATOUVROW_SSSE3
+// TODO: #define HAS_RGBATOYROW_SSSE3
 #define HAS_ARGBTORGBAROW_SSSE3
 #define HAS_ARGB1555TOARGBROW_SSE2
 #define HAS_ARGB4444TOARGBROW_SSE2
@@ -56,11 +58,12 @@ extern "C" {
 #define HAS_BGRATOYROW_SSSE3
 #define HAS_COPYROW_SSE2
 #define HAS_COPYROW_X86
-#define HAS_I444TOARGBROW_SSSE3
 #define HAS_I422TOARGBROW_SSSE3
-#define HAS_I411TOARGBROW_SSSE3
 #define HAS_I422TOBGRAROW_SSSE3
 #define HAS_I422TOABGRROW_SSSE3
+// TODO: #define HAS_I422TORGBAROW_SSSE3
+#define HAS_I444TOARGBROW_SSSE3
+#define HAS_I411TOARGBROW_SSSE3
 #define HAS_I400TOARGBROW_SSE2
 #define HAS_MIRRORROW_SSSE3
 #define HAS_MIRRORROWUV_SSSE3
@@ -112,6 +115,7 @@ extern "C" {
 #define HAS_I422TOARGBROW_NEON
 #define HAS_I422TOBGRAROW_NEON
 #define HAS_I422TOABGRROW_NEON
+// TODO: #define HAS_I422TORGBAROW_NEON
 #endif
 
 #if defined(_MSC_VER) && !defined(__CLR_VER)
@@ -161,13 +165,20 @@ void I422ToABGRRow_NEON(const uint8* y_buf,
                         const uint8* v_buf,
                         uint8* rgb_buf,
                         int width);
+void I422ToRGBARow_NEON(const uint8* y_buf,
+                        const uint8* u_buf,
+                        const uint8* v_buf,
+                        uint8* rgb_buf,
+                        int width);
 
 void ARGBToYRow_SSSE3(const uint8* src_argb, uint8* dst_y, int pix);
 void BGRAToYRow_SSSE3(const uint8* src_argb, uint8* dst_y, int pix);
 void ABGRToYRow_SSSE3(const uint8* src_argb, uint8* dst_y, int pix);
+void RGBAToYRow_SSSE3(const uint8* src_argb, uint8* dst_y, int pix);
 void ARGBToYRow_Unaligned_SSSE3(const uint8* src_argb, uint8* dst_y, int pix);
 void BGRAToYRow_Unaligned_SSSE3(const uint8* src_argb, uint8* dst_y, int pix);
 void ABGRToYRow_Unaligned_SSSE3(const uint8* src_argb, uint8* dst_y, int pix);
+void RGBAToYRow_Unaligned_SSSE3(const uint8* src_argb, uint8* dst_y, int pix);
 
 void ARGBToUVRow_SSSE3(const uint8* src_argb0, int src_stride_argb,
                        uint8* dst_u, uint8* dst_v, int width);
@@ -175,12 +186,16 @@ void BGRAToUVRow_SSSE3(const uint8* src_argb0, int src_stride_argb,
                        uint8* dst_u, uint8* dst_v, int width);
 void ABGRToUVRow_SSSE3(const uint8* src_argb0, int src_stride_argb,
                        uint8* dst_u, uint8* dst_v, int width);
+void RGBAToUVRow_SSSE3(const uint8* src_argb0, int src_stride_argb,
+                       uint8* dst_u, uint8* dst_v, int width);
 void ARGBToUVRow_Unaligned_SSSE3(const uint8* src_argb0, int src_stride_argb,
                                  uint8* dst_u, uint8* dst_v, int width);
 void BGRAToUVRow_Unaligned_SSSE3(const uint8* src_argb0, int src_stride_argb,
                                  uint8* dst_u, uint8* dst_v, int width);
 void ABGRToUVRow_Unaligned_SSSE3(const uint8* src_argb0, int src_stride_argb,
                                  uint8* dst_u, uint8* dst_v, int width);
+void RGBAToUVRow_Unaligned_SSSE3(const uint8* src_argb0, int src_stride_argb,
+                                 uint8* dst_u, uint8* dst_v, int width);
 
 void MirrorRow_SSSE3(const uint8* src, uint8* dst, int width);
 void MirrorRow_SSE2(const uint8* src, uint8* dst, int width);
@@ -206,6 +221,7 @@ void CopyRow_C(const uint8* src, uint8* dst, int count);
 void ARGBToYRow_C(const uint8* src_argb, uint8* dst_y, int pix);
 void BGRAToYRow_C(const uint8* src_argb, uint8* dst_y, int pix);
 void ABGRToYRow_C(const uint8* src_argb, uint8* dst_y, int pix);
+void RGBAToYRow_C(const uint8* src_argb, uint8* dst_y, int pix);
 
 void ARGBToUVRow_C(const uint8* src_argb0, int src_stride_argb,
                    uint8* dst_u, uint8* dst_v, int width);
@@ -213,19 +229,21 @@ void BGRAToUVRow_C(const uint8* src_argb0, int src_stride_argb,
                    uint8* dst_u, uint8* dst_v, int width);
 void ABGRToUVRow_C(const uint8* src_argb0, int src_stride_argb,
                    uint8* dst_u, uint8* dst_v, int width);
+void RGBAToUVRow_C(const uint8* src_argb0, int src_stride_argb,
+                   uint8* dst_u, uint8* dst_v, int width);
 
-void ABGRToARGBRow_SSSE3(const uint8* src_abgr, uint8* dst_argb, int pix);
 void BGRAToARGBRow_SSSE3(const uint8* src_bgra, uint8* dst_argb, int pix);
-void RGBAToARGBRow_SSSE3(const uint8* src_abgr, uint8* dst_argb, int pix);
+void ABGRToARGBRow_SSSE3(const uint8* src_abgr, uint8* dst_argb, int pix);
+void RGBAToARGBRow_SSSE3(const uint8* src_rgba, uint8* dst_argb, int pix);
 void RGB24ToARGBRow_SSSE3(const uint8* src_rgb24, uint8* dst_argb, int pix);
 void RAWToARGBRow_SSSE3(const uint8* src_rgb24, uint8* dst_argb, int pix);
 void ARGB1555ToARGBRow_SSE2(const uint8* src_argb, uint8* dst_argb, int pix);
 void RGB565ToARGBRow_SSE2(const uint8* src_argb, uint8* dst_argb, int pix);
 void ARGB4444ToARGBRow_SSE2(const uint8* src_argb, uint8* dst_argb, int pix);
 
-void ABGRToARGBRow_C(const uint8* src_abgr, uint8* dst_argb, int pix);
 void BGRAToARGBRow_C(const uint8* src_bgra, uint8* dst_argb, int pix);
-void RGBAToARGBRow_C(const uint8* src_abgr, uint8* dst_argb, int pix);
+void ABGRToARGBRow_C(const uint8* src_abgr, uint8* dst_argb, int pix);
+void RGBAToARGBRow_C(const uint8* src_rgba, uint8* dst_argb, int pix);
 void RGB24ToARGBRow_C(const uint8* src_rgb24, uint8* dst_argb, int pix);
 void RAWToARGBRow_C(const uint8* src_rgb24, uint8* dst_argb, int pix);
 void RGB565ToARGBRow_C(const uint8* src_rgb, uint8* dst_argb, int pix);
@@ -289,6 +307,12 @@ void I422ToABGRRow_C(const uint8* y_buf,
                      uint8* abgr_buf,
                      int width);
 
+void I422ToRGBARow_C(const uint8* y_buf,
+                     const uint8* u_buf,
+                     const uint8* v_buf,
+                     uint8* rgba_buf,
+                     int width);
+
 void YToARGBRow_C(const uint8* y_buf,
                   uint8* rgb_buf,
                   int width);
@@ -333,6 +357,12 @@ void I422ToABGRRow_SSSE3(const uint8* y_buf,
                          uint8* abgr_buf,
                          int width);
 
+void I422ToRGBARow_SSSE3(const uint8* y_buf,
+                         const uint8* u_buf,
+                         const uint8* v_buf,
+                         uint8* rgba_buf,
+                         int width);
+
 void I444ToARGBRow_Unaligned_SSSE3(const uint8* y_buf,
                                    const uint8* u_buf,
                                    const uint8* v_buf,
@@ -373,6 +403,12 @@ void I422ToABGRRow_Unaligned_SSSE3(const uint8* y_buf,
                                    uint8* abgr_buf,
                                    int width);
 
+void I422ToRGBARow_Unaligned_SSSE3(const uint8* y_buf,
+                                   const uint8* u_buf,
+                                   const uint8* v_buf,
+                                   uint8* rgba_buf,
+                                   int width);
+
 void I444ToARGBRow_Any_SSSE3(const uint8* y_buf,
                              const uint8* u_buf,
                              const uint8* v_buf,
@@ -413,6 +449,12 @@ void I422ToABGRRow_Any_SSSE3(const uint8* y_buf,
                              uint8* abgr_buf,
                              int width);
 
+void I422ToRGBARow_Any_SSSE3(const uint8* y_buf,
+                             const uint8* u_buf,
+                             const uint8* v_buf,
+                             uint8* rgba_buf,
+                             int width);
+
 void YToARGBRow_SSE2(const uint8* y_buf,
                      uint8* argb_buf,
                      int width);
@@ -434,12 +476,15 @@ void ARGBToARGB4444Row_Any_SSE2(const uint8* src_argb, uint8* dst_rgb, int pix);
 void ARGBToYRow_Any_SSSE3(const uint8* src_argb, uint8* dst_y, int pix);
 void BGRAToYRow_Any_SSSE3(const uint8* src_argb, uint8* dst_y, int pix);
 void ABGRToYRow_Any_SSSE3(const uint8* src_argb, uint8* dst_y, int pix);
+void RGBAToYRow_Any_SSSE3(const uint8* src_argb, uint8* dst_y, int pix);
 void ARGBToUVRow_Any_SSSE3(const uint8* src_argb0, int src_stride_argb,
                            uint8* dst_u, uint8* dst_v, int width);
 void BGRAToUVRow_Any_SSSE3(const uint8* src_argb0, int src_stride_argb,
                            uint8* dst_u, uint8* dst_v, int width);
 void ABGRToUVRow_Any_SSSE3(const uint8* src_argb0, int src_stride_argb,
                            uint8* dst_u, uint8* dst_v, int width);
+void RGBAToUVRow_Any_SSSE3(const uint8* src_argb0, int src_stride_argb,
+                           uint8* dst_u, uint8* dst_v, int width);
 
 void I422ToARGBRow_Any_NEON(const uint8* y_buf,
                             const uint8* u_buf,
@@ -459,6 +504,12 @@ void I422ToABGRRow_Any_NEON(const uint8* y_buf,
                             uint8* rgb_buf,
                             int width);
 
+void I422ToRGBARow_Any_NEON(const uint8* y_buf,
+                            const uint8* u_buf,
+                            const uint8* v_buf,
+                            uint8* rgb_buf,
+                            int width);
+
 void YUY2ToYRow_SSE2(const uint8* src_yuy2, uint8* dst_y, int pix);
 void YUY2ToUVRow_SSE2(const uint8* src_yuy2, int stride_yuy2,
                       uint8* dst_u, uint8* dst_v, int pix);

source/row_common.cc

@@ -19,22 +19,6 @@ namespace libyuv {
 extern "C" {
 #endif
 
-void ABGRToARGBRow_C(const uint8* src_abgr, uint8* dst_argb, int width) {
-  for (int x = 0; x < width; ++x) {
-    // To support in-place conversion.
-    uint8 r = src_abgr[0];
-    uint8 g = src_abgr[1];
-    uint8 b = src_abgr[2];
-    uint8 a = src_abgr[3];
-    dst_argb[0] = b;
-    dst_argb[1] = g;
-    dst_argb[2] = r;
-    dst_argb[3] = a;
-    dst_argb += 4;
-    src_abgr += 4;
-  }
-}
-
 void BGRAToARGBRow_C(const uint8* src_bgra, uint8* dst_argb, int width) {
   for (int x = 0; x < width; ++x) {
     // To support in-place conversion.
@@ -51,6 +35,22 @@ void BGRAToARGBRow_C(const uint8* src_bgra, uint8* dst_argb, int width) {
   }
 }
 
+void ABGRToARGBRow_C(const uint8* src_abgr, uint8* dst_argb, int width) {
+  for (int x = 0; x < width; ++x) {
+    // To support in-place conversion.
+    uint8 r = src_abgr[0];
+    uint8 g = src_abgr[1];
+    uint8 b = src_abgr[2];
+    uint8 a = src_abgr[3];
+    dst_argb[0] = b;
+    dst_argb[1] = g;
+    dst_argb[2] = r;
+    dst_argb[3] = a;
+    dst_argb += 4;
+    src_abgr += 4;
+  }
+}
+
 void RGBAToARGBRow_C(const uint8* src_abgr, uint8* dst_argb, int width) {
   for (int x = 0; x < width; ++x) {
     // To support in-place conversion.
@@ -302,6 +302,7 @@ void NAME ## ToUVRow_C(const uint8* src_rgb0, int src_stride_rgb,              \
 MAKEROWY(ARGB, 2, 1, 0)
 MAKEROWY(BGRA, 1, 2, 3)
 MAKEROWY(ABGR, 0, 1, 2)
+MAKEROWY(RGBA, 3, 2, 1)
 
 // http://en.wikipedia.org/wiki/Grayscale.
 // 0.11 * B + 0.59 * G + 0.30 * R
@@ -574,7 +575,7 @@ void I422ToABGRRow_C(const uint8* y_buf,
                      int width) {
   for (int x = 0; x < width - 1; x += 2) {
     YuvPixel(y_buf[0], u_buf[0], v_buf[0], rgb_buf + 0, 24, 0, 8, 16);
-    YuvPixel(y_buf[1], u_buf[0], v_buf[0], rgb_buf + 4, 24, 0, 8, 16);
+    YuvPixel(y_buf[1], u_buf[0], v_buf[0], rgb_buf + 4, 0, 8, 16, 24);
     y_buf += 2;
     u_buf += 1;
     v_buf += 1;
@@ -585,6 +586,24 @@ void I422ToABGRRow_C(const uint8* y_buf,
   }
 }
 
+void I422ToRGBARow_C(const uint8* y_buf,
+                     const uint8* u_buf,
+                     const uint8* v_buf,
+                     uint8* rgb_buf,
+                     int width) {
+  for (int x = 0; x < width - 1; x += 2) {
+    YuvPixel(y_buf[0], u_buf[0], v_buf[0], rgb_buf + 0, 0, 24, 16, 8);
+    YuvPixel(y_buf[1], u_buf[0], v_buf[0], rgb_buf + 4, 0, 24, 16, 8);
+    y_buf += 2;
+    u_buf += 1;
+    v_buf += 1;
+    rgb_buf += 8;  // Advance 2 pixels.
+  }
+  if (width & 1) {
+    YuvPixel(y_buf[0], u_buf[0], v_buf[0], rgb_buf + 0, 0, 24, 16, 8);
+  }
+}
+
 void YToARGBRow_C(const uint8* y_buf, uint8* rgb_buf, int width) {
   for (int x = 0; x < width; ++x) {
     YuvPixel(y_buf[0], 128, 128, rgb_buf, 24, 16, 8, 0);
@@ -930,11 +949,13 @@ Y2NY(NV12ToARGBRow_Any_SSSE3, NV12ToARGBRow_Unaligned_SSSE3, NV12ToARGBRow_C, 0)
 Y2NY(NV21ToARGBRow_Any_SSSE3, NV21ToARGBRow_Unaligned_SSSE3, NV21ToARGBRow_C, 0)
 YANY(I422ToBGRARow_Any_SSSE3, I422ToBGRARow_Unaligned_SSSE3, I422ToBGRARow_C, 1)
 YANY(I422ToABGRRow_Any_SSSE3, I422ToABGRRow_Unaligned_SSSE3, I422ToABGRRow_C, 1)
+// TODO: YANY(I422ToRGBARow_Any_SSSE3, I422ToRGBARow_Unaligned_SSSE3, I422ToRGBARow_C, 1)
 #endif
 #if defined(HAS_I422TOARGBROW_NEON)
 YANY(I422ToARGBRow_Any_NEON, I422ToARGBRow_NEON, I422ToARGBRow_C, 1)
 YANY(I422ToBGRARow_Any_NEON, I422ToBGRARow_NEON, I422ToBGRARow_C, 1)
 YANY(I422ToABGRRow_Any_NEON, I422ToABGRRow_NEON, I422ToABGRRow_C, 1)
+// TODO: YANY(I422ToRGBARow_Any_NEON, I422ToRGBARow_NEON, I422ToRGBARow_C, 1)
 #endif
 #undef YANY
 
@@ -966,6 +987,7 @@ RGBANY(ARGBToARGB4444Row_Any_SSE2, ARGBToARGB4444Row_SSE2, 2)
 YANY(ARGBToYRow_Any_SSSE3, ARGBToYRow_Unaligned_SSSE3, 4)
 YANY(BGRAToYRow_Any_SSSE3, BGRAToYRow_Unaligned_SSSE3, 4)
 YANY(ABGRToYRow_Any_SSSE3, ABGRToYRow_Unaligned_SSSE3, 4)
+// TODO: YANY(RGBAToYRow_Any_SSSE3, RGBAToYRow_Unaligned_SSSE3, 4)
 YANY(YUY2ToYRow_Any_SSE2, YUY2ToYRow_Unaligned_SSE2, 2)
 YANY(UYVYToYRow_Any_SSE2, UYVYToYRow_Unaligned_SSE2, 2)
 #undef YANY
@@ -984,6 +1006,7 @@ YANY(UYVYToYRow_Any_SSE2, UYVYToYRow_Unaligned_SSE2, 2)
 UVANY(ARGBToUVRow_Any_SSSE3, ARGBToUVRow_Unaligned_SSSE3, ARGBToUVRow_C, 4)
 UVANY(BGRAToUVRow_Any_SSSE3, BGRAToUVRow_Unaligned_SSSE3, BGRAToUVRow_C, 4)
 UVANY(ABGRToUVRow_Any_SSSE3, ABGRToUVRow_Unaligned_SSSE3, ABGRToUVRow_C, 4)
+// TODO: UVANY(RGBAToUVRow_Any_SSSE3, RGBAToUVRow_Unaligned_SSSE3, RGBAToUVRow_C, 4)
 UVANY(YUY2ToUVRow_Any_SSE2, YUY2ToUVRow_Unaligned_SSE2, YUY2ToUVRow_C, 2)
 UVANY(UYVYToUVRow_Any_SSE2, UYVYToUVRow_Unaligned_SSE2, UYVYToUVRow_C, 2)
 #undef UVANY

unit_test/planar_test.cc

@@ -86,6 +86,7 @@ TEST_F(libyuvTest, FMT_PLANAR##To##FMT_B##N##_OptVsC) {                        \
 TESTPLANARTOB(I420, 2, 2, ARGB, 4)
 TESTPLANARTOB(I420, 2, 2, BGRA, 4)
 TESTPLANARTOB(I420, 2, 2, ABGR, 4)
+TESTPLANARTOB(I420, 2, 2, RGBA, 4)
 TESTPLANARTOB(I420, 2, 2, RAW, 3)
 TESTPLANARTOB(I420, 2, 2, RGB24, 3)
 TESTPLANARTOB(I420, 2, 2, RGB565, 2)
@@ -239,6 +240,7 @@ TEST_F(libyuvTest, FMT_A##To##FMT_PLANAR##N##_OptVsC) {                        \
 TESTATOPLANAR(ARGB, 4, I420, 2, 2)
 TESTATOPLANAR(BGRA, 4, I420, 2, 2)
 TESTATOPLANAR(ABGR, 4, I420, 2, 2)
+TESTATOPLANAR(RGBA, 4, I420, 2, 2)
 TESTATOPLANAR(RAW, 3, I420, 2, 2)
 TESTATOPLANAR(RGB24, 3, I420, 2, 2)
 TESTATOPLANAR(RGB565, 2, I420, 2, 2)

unit_test/unit_test.cc

@@ -10,10 +10,10 @@
 
 #include "unit_test/unit_test.h"
 
-#include <cstring>
-
 #include <stdlib.h>  // For getenv()
 
+#include <cstring>
+
 // Change this to 1000 for benchmarking.
 // TODO(fbarchard): Add command line parsing to pass this as option.
 #define BENCHMARK_ITERATIONS 1
@@ -23,7 +23,7 @@ libyuvTest::libyuvTest() : rotate_max_w_(128), rotate_max_h_(128),
     benchmark_height_(720) {
     const char* repeat = getenv("LIBYUV_REPEAT");
     if (repeat) {
-      benchmark_iterations_ = atoi(repeat);
+      benchmark_iterations_ = atoi(repeat);  // NOLINT
     }
 }