yuvconstants for all YUV to RGB conversion functions.

source/row_win.cc

 /*
  *  Copyright 2011 The LibYuv Project Authors. All rights reserved.
  *
  *  Use of this source code is governed by a BSD-style license
  *  that can be found in the LICENSE file in the root of the source
  *  tree. An additional intellectual property rights grant can be found
  *  in the file PATENTS. All contributing project authors may
  *  be found in the AUTHORS file in the root of the source tree.
  */
 
@@ skipping 65 matching lines @@
     xmm2 = _mm_unpacklo_epi8(xmm2, xmm1);                                      \
     xmm0 = _mm_unpacklo_epi8(xmm0, xmm5);                                      \
     xmm1 = _mm_loadu_si128(&xmm2);                                             \
     xmm2 = _mm_unpacklo_epi16(xmm2, xmm0);                                     \
     xmm1 = _mm_unpackhi_epi16(xmm1, xmm0);                                     \
     _mm_storeu_si128((__m128i *)dst_argb, xmm2);                               \
     _mm_storeu_si128((__m128i *)(dst_argb + 16), xmm1);                        \
     dst_argb += 32;
 
 
-#if defined(HAS_I422TOARGBMATRIXROW_SSSE3)
-void I422ToARGBMatrixRow_SSSE3(const uint8* y_buf,
-                               const uint8* u_buf,
-                               const uint8* v_buf,
-                               uint8* dst_argb,
-                               struct YuvConstants* YuvConstants,
-                               int width) {
+#if defined(HAS_I422TOARGBROW_SSSE3)
+void I422ToARGBRow_SSSE3(const uint8* y_buf,
+                         const uint8* u_buf,
+                         const uint8* v_buf,
+                         uint8* dst_argb,
+                         struct YuvConstants* yuvconstants,
+                         int width) {
   __m128i xmm0, xmm1, xmm2, xmm3;
   const __m128i xmm5 = _mm_set1_epi8(-1);
   const ptrdiff_t offset = (uint8*)v_buf - (uint8*)u_buf;
   while (width > 0) {
     READYUV422
     YUVTORGB(YuvConstants)
     STOREARGB
     width -= 8;
   }
 }
 #endif
 
-#if defined(HAS_I422TOABGRMATRIXROW_SSSE3)
-void I422ToABGRMatrixRow_SSSE3(const uint8* y_buf,
-                               const uint8* u_buf,
-                               const uint8* v_buf,
-                               uint8* dst_argb,
-                               struct YuvConstants* YuvConstants,
-                               int width) {
+#if defined(HAS_I422TOABGRROW_SSSE3)
+void I422ToABGRRow_SSSE3(const uint8* y_buf,
+                         const uint8* u_buf,
+                         const uint8* v_buf,
+                         uint8* dst_argb,
+                         struct YuvConstants* yuvconstants,
+                         int width) {
   __m128i xmm0, xmm1, xmm2, xmm3;
   const __m128i xmm5 = _mm_set1_epi8(-1);
   const ptrdiff_t offset = (uint8*)v_buf - (uint8*)u_buf;
   while (width > 0) {
     READYUV422
     YUVTORGB(YuvConstants)
     STOREABGR
     width -= 8;
   }
 }
@@ skipping 1834 matching lines @@
     __asm vpermq     ymm1, ymm1, 0xd8                                          \
     __asm vpunpcklbw ymm2, ymm0, ymm5           /* BA */                       \
     __asm vpermq     ymm2, ymm2, 0xd8                                          \
     __asm vpunpcklwd ymm0, ymm1, ymm2           /* RGBA first 8 pixels */      \
     __asm vpunpckhwd ymm1, ymm1, ymm2           /* RGBA next 8 pixels */       \
     __asm vmovdqu    [edx], ymm0                                               \
     __asm vmovdqu    [edx + 32], ymm1                                          \
     __asm lea        edx,  [edx + 64]                                          \
   }
 
-#ifdef HAS_I422TOARGBMATRIXROW_AVX2
+#ifdef HAS_I422TOARGBROW_AVX2
 // 16 pixels
 // 8 UV values upsampled to 16 UV, mixed with 16 Y producing 16 ARGB (64 bytes).
 __declspec(naked)
-void I422ToARGBMatrixRow_AVX2(const uint8* y_buf,
-                              const uint8* u_buf,
-                              const uint8* v_buf,
-                              uint8* dst_argb,
-                              struct YuvConstants* YuvConstants,
-                              int width) {
+void I422ToARGBRow_AVX2(const uint8* y_buf,
+                        const uint8* u_buf,
+                        const uint8* v_buf,
+                        uint8* dst_argb,
+                        struct YuvConstants* yuvconstants,
+                        int width) {
   __asm {
     push       esi
     push       edi
     push       ebp
     mov        eax, [esp + 12 + 4]   // Y
     mov        esi, [esp + 12 + 8]   // U
     mov        edi, [esp + 12 + 12]  // V
     mov        edx, [esp + 12 + 16]  // argb
     mov        ebp, [esp + 12 + 20]  // YuvConstants
     mov        ecx, [esp + 12 + 24]  // width
     sub        edi, esi
     vpcmpeqb   ymm5, ymm5, ymm5     // generate 0xffffffffffffffff for alpha
 
  convertloop:
     READYUV422_AVX2
     YUVTORGB_AVX2(ebp)
     STOREARGB_AVX2
 
     sub        ecx, 16
     jg         convertloop
 
     pop        ebp
     pop        edi
     pop        esi
     vzeroupper
     ret
   }
 }
-#endif  // HAS_I422TOARGBMATRIXROW_AVX2
+#endif  // HAS_I422TOARGBROW_AVX2
 
-#ifdef HAS_I444TOARGBMATRIXROW_AVX2
+#ifdef HAS_I444TOARGBROW_AVX2
 // 16 pixels
 // 16 UV values with 16 Y producing 16 ARGB (64 bytes).
 __declspec(naked)
-void I444ToARGBMatrixRow_AVX2(const uint8* y_buf,
-                              const uint8* u_buf,
-                              const uint8* v_buf,
-                              uint8* dst_argb,
-                              struct YuvConstants* YuvConstants,
-                              int width) {
+void I444ToARGBRow_AVX2(const uint8* y_buf,
+                        const uint8* u_buf,
+                        const uint8* v_buf,
+                        uint8* dst_argb,
+                        struct YuvConstants* yuvconstants,
+                        int width) {
   __asm {
     push       esi
     push       edi
     push       ebp
     mov        eax, [esp + 12 + 4]   // Y
     mov        esi, [esp + 12 + 8]   // U
     mov        edi, [esp + 12 + 12]  // V
     mov        edx, [esp + 12 + 16]  // argb
     mov        ebp, [esp + 12 + 20]  // YuvConstants
     mov        ecx, [esp + 12 + 24]  // width
     sub        edi, esi
     vpcmpeqb   ymm5, ymm5, ymm5     // generate 0xffffffffffffffff for alpha
  convertloop:
     READYUV444_AVX2
     YUVTORGB_AVX2(ebp)
     STOREARGB_AVX2
 
     sub        ecx, 16
     jg         convertloop
 
     pop        ebp
     pop        edi
     pop        esi
     vzeroupper
     ret
   }
 }
-#endif  // HAS_I444TOARGBMATRIXROW_AVX2
+#endif  // HAS_I444TOARGBROW_AVX2
 
-#ifdef HAS_I444TOABGRMATRIXROW_AVX2
+#ifdef HAS_I444TOABGRROW_AVX2
 // 16 pixels
 // 16 UV values with 16 Y producing 16 ABGR (64 bytes).
 __declspec(naked)
-void I444ToABGRMatrixRow_AVX2(const uint8* y_buf,
-                              const uint8* u_buf,
-                              const uint8* v_buf,
-                              uint8* dst_abgr,
-                              struct YuvConstants* YuvConstants,
-                              int width) {
+void I444ToABGRRow_AVX2(const uint8* y_buf,
+                        const uint8* u_buf,
+                        const uint8* v_buf,
+                        uint8* dst_abgr,
+                        struct YuvConstants* yuvconstants,
+                        int width) {
   __asm {
     push       esi
     push       edi
     push       ebp
     mov        eax, [esp + 12 + 4]   // Y
     mov        esi, [esp + 12 + 8]   // U
     mov        edi, [esp + 12 + 12]  // V
     mov        edx, [esp + 12 + 16]  // abgr
     mov        ebp, [esp + 12 + 20]  // YuvConstants
     mov        ecx, [esp + 12 + 24]  // width
     sub        edi, esi
     vpcmpeqb   ymm5, ymm5, ymm5     // generate 0xffffffffffffffff for alpha
  convertloop:
     READYUV444_AVX2
     YUVTORGB_AVX2(ebp)
     STOREABGR_AVX2
 
     sub        ecx, 16
     jg         convertloop
 
     pop        ebp
     pop        edi
     pop        esi
     vzeroupper
     ret
   }
 }
-#endif  // HAS_I444TOABGRMATRIXROW_AVX2
+#endif  // HAS_I444TOABGRROW_AVX2
 
 #ifdef HAS_I411TOARGBROW_AVX2
 // 16 pixels
 // 4 UV values upsampled to 16 UV, mixed with 16 Y producing 16 ARGB (64 bytes).
 __declspec(naked)
 void I411ToARGBRow_AVX2(const uint8* y_buf,
                         const uint8* u_buf,
                         const uint8* v_buf,
                         uint8* dst_argb,
+                        struct YuvConstants* yuvconstants,
                         int width) {
   __asm {
     push       esi
     push       edi
-    mov        eax, [esp + 8 + 4]   // Y
-    mov        esi, [esp + 8 + 8]   // U
-    mov        edi, [esp + 8 + 12]  // V
-    mov        edx, [esp + 8 + 16]  // argb
-    mov        ecx, [esp + 8 + 20]  // width
+    push       ebp
+    mov        eax, [esp + 12 + 4]   // Y
+    mov        esi, [esp + 12 + 8]   // U
+    mov        edi, [esp + 12 + 12]  // V
+    mov        edx, [esp + 12 + 16]  // abgr
+    mov        ebp, [esp + 12 + 20]  // YuvConstants
+    mov        ecx, [esp + 12 + 24]  // width
     sub        edi, esi
     vpcmpeqb   ymm5, ymm5, ymm5     // generate 0xffffffffffffffff for alpha
 
  convertloop:
     READYUV411_AVX2
-    YUVTORGB_AVX2(kYuvConstants)
+    YUVTORGB_AVX2(ebp)
     STOREARGB_AVX2
 
     sub        ecx, 16
     jg         convertloop
 
+    pop        ebp
     pop        edi
     pop        esi
     vzeroupper
     ret
   }
 }
 #endif  // HAS_I411TOARGBROW_AVX2
 
 #ifdef HAS_NV12TOARGBROW_AVX2
 // 16 pixels.
 // 8 UV values upsampled to 16 UV, mixed with 16 Y producing 16 ARGB (64 bytes).
 __declspec(naked)
 void NV12ToARGBRow_AVX2(const uint8* y_buf,
                         const uint8* uv_buf,
                         uint8* dst_argb,
+                        struct YuvConstants* yuvconstants,
                         int width) {
   __asm {
     push       esi
-    mov        eax, [esp + 4 + 4]   // Y
-    mov        esi, [esp + 4 + 8]   // UV
-    mov        edx, [esp + 4 + 12]  // argb
-    mov        ecx, [esp + 4 + 16]  // width
+    push       ebp
+    mov        eax, [esp + 8 + 4]   // Y
+    mov        esi, [esp + 8 + 8]   // UV
+    mov        edx, [esp + 8 + 12]  // argb
+    mov        ebp, [esp + 8 + 16]  // YuvConstants
+    mov        ecx, [esp + 8 + 20]  // width
     vpcmpeqb   ymm5, ymm5, ymm5     // generate 0xffffffffffffffff for alpha
 
  convertloop:
     READNV12_AVX2
-    YUVTORGB_AVX2(kYuvConstants)
+    YUVTORGB_AVX2(ebp)
     STOREARGB_AVX2
 
     sub        ecx, 16
     jg         convertloop
 
+    pop        ebp
     pop        esi
     vzeroupper
     ret
   }
 }
 #endif  // HAS_NV12TOARGBROW_AVX2
 
-#ifdef HAS_NV21TOARGBROW_AVX2
-// 16 pixels.
-// 8 VU values upsampled to 16 VU, mixed with 16 Y producing 16 ARGB (64 bytes).
-__declspec(naked)
-void NV21ToARGBRow_AVX2(const uint8* y_buf,
-                        const uint8* uv_buf,
-                        uint8* dst_argb,
-                        int width) {
-  __asm {
-    push       esi
-    mov        eax, [esp + 4 + 4]   // Y
-    mov        esi, [esp + 4 + 8]   // UV
-    mov        edx, [esp + 4 + 12]  // argb
-    mov        ecx, [esp + 4 + 16]  // width
-    vpcmpeqb   ymm5, ymm5, ymm5     // generate 0xffffffffffffffff for alpha
-
- convertloop:
-    READNV12_AVX2
-    YUVTORGB_AVX2(kYvuConstants)
-    STOREARGB_AVX2
-
-    sub        ecx, 16
-    jg         convertloop
-
-    pop        esi
-    vzeroupper
-    ret
-  }
-}
-#endif  // HAS_NV21TOARGBROW_AVX2
-
 #ifdef HAS_I422TOBGRAROW_AVX2
 // 16 pixels
 // 8 UV values upsampled to 16 UV, mixed with 16 Y producing 16 BGRA (64 bytes).
 // TODO(fbarchard): Use macros to reduce duplicate code.  See SSSE3.
 __declspec(naked)
 void I422ToBGRARow_AVX2(const uint8* y_buf,
                         const uint8* u_buf,
                         const uint8* v_buf,
                         uint8* dst_argb,
+                        struct YuvConstants* yuvconstants,
                         int width) {
   __asm {
     push       esi
     push       edi
-    mov        eax, [esp + 8 + 4]   // Y
-    mov        esi, [esp + 8 + 8]   // U
-    mov        edi, [esp + 8 + 12]  // V
-    mov        edx, [esp + 8 + 16]  // argb
-    mov        ecx, [esp + 8 + 20]  // width
+    push       ebp
+    mov        eax, [esp + 12 + 4]   // Y
+    mov        esi, [esp + 12 + 8]   // U
+    mov        edi, [esp + 12 + 12]  // V
+    mov        edx, [esp + 12 + 16]  // abgr
+    mov        ebp, [esp + 12 + 20]  // YuvConstants
+    mov        ecx, [esp + 12 + 24]  // width
     sub        edi, esi
     vpcmpeqb   ymm5, ymm5, ymm5     // generate 0xffffffffffffffff for alpha
 
  convertloop:
     READYUV422_AVX2
-    YUVTORGB_AVX2(kYuvConstants)
+    YUVTORGB_AVX2(ebp)
     STOREBGRA_AVX2
 
     sub        ecx, 16
     jg         convertloop
 
+    pop        ebp
     pop        edi
     pop        esi
     vzeroupper
     ret
   }
 }
 #endif  // HAS_I422TOBGRAROW_AVX2
 
 #ifdef HAS_I422TORGBAROW_AVX2
 // 16 pixels
 // 8 UV values upsampled to 16 UV, mixed with 16 Y producing 16 RGBA (64 bytes).
 __declspec(naked)
 void I422ToRGBARow_AVX2(const uint8* y_buf,
                         const uint8* u_buf,
                         const uint8* v_buf,
                         uint8* dst_argb,
+                        struct YuvConstants* yuvconstants,
                         int width) {
   __asm {
     push       esi
     push       edi
-    mov        eax, [esp + 8 + 4]   // Y
-    mov        esi, [esp + 8 + 8]   // U
-    mov        edi, [esp + 8 + 12]  // V
-    mov        edx, [esp + 8 + 16]  // argb
-    mov        ecx, [esp + 8 + 20]  // width
+    push       ebp
+    mov        eax, [esp + 12 + 4]   // Y
+    mov        esi, [esp + 12 + 8]   // U
+    mov        edi, [esp + 12 + 12]  // V
+    mov        edx, [esp + 12 + 16]  // abgr
+    mov        ebp, [esp + 12 + 20]  // YuvConstants
+    mov        ecx, [esp + 12 + 24]  // width
     sub        edi, esi
     vpcmpeqb   ymm5, ymm5, ymm5     // generate 0xffffffffffffffff for alpha
 
  convertloop:
     READYUV422_AVX2
-    YUVTORGB_AVX2(kYuvConstants)
+    YUVTORGB_AVX2(ebp)
     STORERGBA_AVX2
 
     sub        ecx, 16
     jg         convertloop
 
+    pop        ebp
     pop        edi
     pop        esi
     vzeroupper
     ret
   }
 }
 #endif  // HAS_I422TORGBAROW_AVX2
 
 #ifdef HAS_I422TOABGRROW_AVX2
 // 16 pixels
 // 8 UV values upsampled to 16 UV, mixed with 16 Y producing 16 ABGR (64 bytes).
 __declspec(naked)
-void I422ToABGRMatrixRow_AVX2(const uint8* y_buf,
-                              const uint8* u_buf,
-                              const uint8* v_buf,
-                              uint8* dst_argb,
-                              struct YuvConstants* YuvConstants,
-                              int width) {
+void I422ToABGRRow_AVX2(const uint8* y_buf,
+                        const uint8* u_buf,
+                        const uint8* v_buf,
+                        uint8* dst_argb,
+                        struct YuvConstants* yuvconstants,
+                        int width) {
   __asm {
     push       esi
     push       edi
     push       ebp
     mov        eax, [esp + 12 + 4]   // Y
     mov        esi, [esp + 12 + 8]   // U
     mov        edi, [esp + 12 + 12]  // V
     mov        edx, [esp + 12 + 16]  // argb
     mov        ebp, [esp + 12 + 20]  // YuvConstants
     mov        ecx, [esp + 12 + 24]  // width
@@ skipping 199 matching lines @@
     __asm por        xmm3, xmm2    /* BG */                                    \
     __asm por        xmm1, xmm3    /* BGR */                                   \
     __asm packssdw   xmm0, xmm1                                                \
     __asm movdqu     0[edx], xmm0  /* store 8 pixels of RGB565 */              \
     __asm lea        edx, [edx + 16]                                           \
   }
 
 // 8 pixels.
 // 8 UV values, mixed with 8 Y producing 8 ARGB (32 bytes).
 __declspec(naked)
-void I444ToARGBMatrixRow_SSSE3(const uint8* y_buf,
-                               const uint8* u_buf,
-                               const uint8* v_buf,
-                               uint8* dst_argb,
-                               struct YuvConstants* YuvConstants,
-                               int width) {
+void I444ToARGBRow_SSSE3(const uint8* y_buf,
+                         const uint8* u_buf,
+                         const uint8* v_buf,
+                         uint8* dst_argb,
+                         struct YuvConstants* yuvconstants,
+                         int width) {
   __asm {
     push       esi
     push       edi
     push       ebp
     mov        eax, [esp + 12 + 4]   // Y
     mov        esi, [esp + 12 + 8]   // U
     mov        edi, [esp + 12 + 12]  // V
     mov        edx, [esp + 12 + 16]  // argb
     mov        ebp, [esp + 12 + 20]  // YuvConstants
     mov        ecx, [esp + 12 + 24]  // width
@@ skipping 11 matching lines @@
     pop        ebp
     pop        edi
     pop        esi
     ret
   }
 }
 
 // 8 pixels.
 // 8 UV values, mixed with 8 Y producing 8 ABGR (32 bytes).
 __declspec(naked)
-void I444ToABGRMatrixRow_SSSE3(const uint8* y_buf,
-                               const uint8* u_buf,
-                               const uint8* v_buf,
-                               uint8* dst_abgr,
-                               struct YuvConstants* YuvConstants,
-                               int width) {
+void I444ToABGRRow_SSSE3(const uint8* y_buf,
+                         const uint8* u_buf,
+                         const uint8* v_buf,
+                         uint8* dst_abgr,
+                         struct YuvConstants* yuvconstants,
+                         int width) {
   __asm {
     push       esi
     push       edi
     push       ebp
     mov        eax, [esp + 12 + 4]   // Y
     mov        esi, [esp + 12 + 8]   // U
     mov        edi, [esp + 12 + 12]  // V
     mov        edx, [esp + 12 + 16]  // abgr
     mov        ebp, [esp + 12 + 20]  // YuvConstants
     mov        ecx, [esp + 12 + 24]  // width
@@ skipping 15 matching lines @@
   }
 }
 
 // 8 pixels.
 // 4 UV values upsampled to 8 UV, mixed with 8 Y producing 8 RGB24 (24 bytes).
 __declspec(naked)
 void I422ToRGB24Row_SSSE3(const uint8* y_buf,
                           const uint8* u_buf,
                           const uint8* v_buf,
                           uint8* dst_rgb24,
+                          struct YuvConstants* yuvconstants,
                           int width) {
   __asm {
     push       esi
     push       edi
-    mov        eax, [esp + 8 + 4]   // Y
-    mov        esi, [esp + 8 + 8]   // U
-    mov        edi, [esp + 8 + 12]  // V
-    mov        edx, [esp + 8 + 16]  // rgb24
-    mov        ecx, [esp + 8 + 20]  // width
+    push       ebp
+    mov        eax, [esp + 12 + 4]   // Y
+    mov        esi, [esp + 12 + 8]   // U
+    mov        edi, [esp + 12 + 12]  // V
+    mov        edx, [esp + 12 + 16]  // argb
+    mov        ebp, [esp + 12 + 20]  // YuvConstants
+    mov        ecx, [esp + 12 + 24]  // width
     sub        edi, esi
     movdqa     xmm5, xmmword ptr kShuffleMaskARGBToRGB24_0
     movdqa     xmm6, xmmword ptr kShuffleMaskARGBToRGB24
 
  convertloop:
     READYUV422
-    YUVTORGB(kYuvConstants)
+    YUVTORGB(ebp)
     STORERGB24
 
     sub        ecx, 8
     jg         convertloop
 
+    pop        ebp
     pop        edi
     pop        esi
     ret
   }
 }
 
 // 8 pixels.
 // 4 UV values upsampled to 8 UV, mixed with 8 Y producing 8 RAW (24 bytes).
 __declspec(naked)
 void I422ToRAWRow_SSSE3(const uint8* y_buf,
                         const uint8* u_buf,
                         const uint8* v_buf,
                         uint8* dst_raw,
+                        struct YuvConstants* yuvconstants,
                         int width) {
   __asm {
     push       esi
     push       edi
-    mov        eax, [esp + 8 + 4]   // Y
-    mov        esi, [esp + 8 + 8]   // U
-    mov        edi, [esp + 8 + 12]  // V
-    mov        edx, [esp + 8 + 16]  // raw
-    mov        ecx, [esp + 8 + 20]  // width
+    push       ebp
+    mov        eax, [esp + 12 + 4]   // Y
+    mov        esi, [esp + 12 + 8]   // U
+    mov        edi, [esp + 12 + 12]  // V
+    mov        edx, [esp + 12 + 16]  // argb
+    mov        ebp, [esp + 12 + 20]  // YuvConstants
+    mov        ecx, [esp + 12 + 24]  // width
     sub        edi, esi
     movdqa     xmm5, xmmword ptr kShuffleMaskARGBToRAW_0
     movdqa     xmm6, xmmword ptr kShuffleMaskARGBToRAW
 
  convertloop:
     READYUV422
-    YUVTORGB(kYuvConstants)
+    YUVTORGB(ebp)
     STORERAW
 
     sub        ecx, 8
     jg         convertloop
 
+    pop        ebp
     pop        edi
     pop        esi
     ret
   }
 }
 
 // 8 pixels
 // 4 UV values upsampled to 8 UV, mixed with 8 Y producing 8 RGB565 (16 bytes).
 __declspec(naked)
 void I422ToRGB565Row_SSSE3(const uint8* y_buf,
                            const uint8* u_buf,
                            const uint8* v_buf,
                            uint8* rgb565_buf,
+                           struct YuvConstants* yuvconstants,
                            int width) {
   __asm {
     push       esi
     push       edi
-    mov        eax, [esp + 8 + 4]   // Y
-    mov        esi, [esp + 8 + 8]   // U
-    mov        edi, [esp + 8 + 12]  // V
-    mov        edx, [esp + 8 + 16]  // rgb565
-    mov        ecx, [esp + 8 + 20]  // width
+    push       ebp
+    mov        eax, [esp + 12 + 4]   // Y
+    mov        esi, [esp + 12 + 8]   // U
+    mov        edi, [esp + 12 + 12]  // V
+    mov        edx, [esp + 12 + 16]  // argb
+    mov        ebp, [esp + 12 + 20]  // YuvConstants
+    mov        ecx, [esp + 12 + 24]  // width
     sub        edi, esi
     pcmpeqb    xmm5, xmm5       // generate mask 0x0000001f
     psrld      xmm5, 27
     pcmpeqb    xmm6, xmm6       // generate mask 0x000007e0
     psrld      xmm6, 26
     pslld      xmm6, 5
     pcmpeqb    xmm7, xmm7       // generate mask 0xfffff800
     pslld      xmm7, 11
 
  convertloop:
     READYUV422
-    YUVTORGB(kYuvConstants)
+    YUVTORGB(ebp)
     STORERGB565
 
     sub        ecx, 8
     jg         convertloop
 
+    pop        ebp
     pop        edi
     pop        esi
     ret
   }
 }
 
 // 8 pixels.
 // 4 UV values upsampled to 8 UV, mixed with 8 Y producing 8 ARGB (32 bytes).
 __declspec(naked)
-void I422ToARGBMatrixRow_SSSE3(const uint8* y_buf,
-                               const uint8* u_buf,
-                               const uint8* v_buf,
-                               uint8* dst_argb,
-                               struct YuvConstants* YuvConstants,
-                               int width) {
+void I422ToARGBRow_SSSE3(const uint8* y_buf,
+                         const uint8* u_buf,
+                         const uint8* v_buf,
+                         uint8* dst_argb,
+                         struct YuvConstants* yuvconstants,
+                         int width) {
   __asm {
     push       esi
     push       edi
     push       ebp
     mov        eax, [esp + 12 + 4]   // Y
     mov        esi, [esp + 12 + 8]   // U
     mov        edi, [esp + 12 + 12]  // V
     mov        edx, [esp + 12 + 16]  // argb
     mov        ebp, [esp + 12 + 20]  // YuvConstants
     mov        ecx, [esp + 12 + 24]  // width
@@ skipping 16 matching lines @@
 }
 
 // 8 pixels.
 // 2 UV values upsampled to 8 UV, mixed with 8 Y producing 8 ARGB (32 bytes).
 // Similar to I420 but duplicate UV once more.
 __declspec(naked)
 void I411ToARGBRow_SSSE3(const uint8* y_buf,
                          const uint8* u_buf,
                          const uint8* v_buf,
                          uint8* dst_argb,
+                         struct YuvConstants* yuvconstants,
                          int width) {
   __asm {
-    push       ebx
     push       esi
     push       edi
+    push       ebp
     mov        eax, [esp + 12 + 4]   // Y
     mov        esi, [esp + 12 + 8]   // U
     mov        edi, [esp + 12 + 12]  // V
-    mov        edx, [esp + 12 + 16]  // argb
-    mov        ecx, [esp + 12 + 20]  // width
+    mov        edx, [esp + 12 + 16]  // abgr
+    mov        ebp, [esp + 12 + 20]  // YuvConstants
+    mov        ecx, [esp + 12 + 24]  // width
     sub        edi, esi
     pcmpeqb    xmm5, xmm5            // generate 0xffffffff for alpha
 
  convertloop:
-    READYUV411  // modifies EBX
-    YUVTORGB(kYuvConstants)
+    READYUV411
+    YUVTORGB(ebp)
     STOREARGB
 
     sub        ecx, 8
     jg         convertloop
 
+    pop        ebp
     pop        edi
     pop        esi
-    pop        ebx
     ret
   }
 }
 
 // 8 pixels.
 // 4 UV values upsampled to 8 UV, mixed with 8 Y producing 8 ARGB (32 bytes).
 __declspec(naked)
 void NV12ToARGBRow_SSSE3(const uint8* y_buf,
                          const uint8* uv_buf,
                          uint8* dst_argb,
+                         struct YuvConstants* yuvconstants,
                          int width) {
   __asm {
     push       esi
-    mov        eax, [esp + 4 + 4]   // Y
-    mov        esi, [esp + 4 + 8]   // UV
-    mov        edx, [esp + 4 + 12]  // argb
-    mov        ecx, [esp + 4 + 16]  // width
+    push       ebp
+    mov        eax, [esp + 8 + 4]   // Y
+    mov        esi, [esp + 8 + 8]   // UV
+    mov        edx, [esp + 8 + 12]  // argb
+    mov        ebp, [esp + 8 + 16]  // YuvConstants
+    mov        ecx, [esp + 8 + 20]  // width
     pcmpeqb    xmm5, xmm5           // generate 0xffffffff for alpha
 
  convertloop:
     READNV12
-    YUVTORGB(kYuvConstants)
+    YUVTORGB(ebp)
     STOREARGB
 
     sub        ecx, 8
     jg         convertloop
 
+    pop        ebp
     pop        esi
     ret
   }
-}
-
-// 8 pixels.
-// 4 VU values upsampled to 8 VU, mixed with 8 Y producing 8 ARGB (32 bytes).
-__declspec(naked)
-void NV21ToARGBRow_SSSE3(const uint8* y_buf,
-                         const uint8* uv_buf,
-                         uint8* dst_argb,
-                         int width) {
-  __asm {
-    push       esi
-    mov        eax, [esp + 4 + 4]   // Y
-    mov        esi, [esp + 4 + 8]   // UV
-    mov        edx, [esp + 4 + 12]  // argb
-    mov        ecx, [esp + 4 + 16]  // width
-    pcmpeqb    xmm5, xmm5           // generate 0xffffffff for alpha
-
- convertloop:
-    READNV12
-    YUVTORGB(kYvuConstants)
-    STOREARGB
-
-    sub        ecx, 8
-    jg         convertloop
-
-    pop        esi
-    ret
-  }
 }
 
 __declspec(naked)
 void I422ToBGRARow_SSSE3(const uint8* y_buf,
                          const uint8* u_buf,
                          const uint8* v_buf,
                          uint8* dst_bgra,
+                         struct YuvConstants* yuvconstants,
                          int width) {
   __asm {
     push       esi
     push       edi
-    mov        eax, [esp + 8 + 4]   // Y
-    mov        esi, [esp + 8 + 8]   // U
-    mov        edi, [esp + 8 + 12]  // V
-    mov        edx, [esp + 8 + 16]  // bgra
-    mov        ecx, [esp + 8 + 20]  // width
+    push       ebp
+    mov        eax, [esp + 12 + 4]   // Y
+    mov        esi, [esp + 12 + 8]   // U
+    mov        edi, [esp + 12 + 12]  // V
+    mov        edx, [esp + 12 + 16]  // argb
+    mov        ebp, [esp + 12 + 20]  // YuvConstants
+    mov        ecx, [esp + 12 + 24]  // width
     sub        edi, esi
 
  convertloop:
     READYUV422
-    YUVTORGB(kYuvConstants)
+    YUVTORGB(ebp)
     STOREBGRA
 
     sub        ecx, 8
     jg         convertloop
 
+    pop        ebp
     pop        edi
     pop        esi
     ret
   }
 }
 
 __declspec(naked)
-void I422ToABGRMatrixRow_SSSE3(const uint8* y_buf,
-                               const uint8* u_buf,
-                               const uint8* v_buf,
-                               uint8* dst_abgr,
-                               struct YuvConstants* YuvConstants,
-                               int width) {
+void I422ToABGRRow_SSSE3(const uint8* y_buf,
+                         const uint8* u_buf,
+                         const uint8* v_buf,
+                         uint8* dst_abgr,
+                         struct YuvConstants* yuvconstants,
+                         int width) {
   __asm {
     push       esi
     push       edi
     push       ebp
     mov        eax, [esp + 12 + 4]   // Y
     mov        esi, [esp + 12 + 8]   // U
     mov        edi, [esp + 12 + 12]  // V
     mov        edx, [esp + 12 + 16]  // argb
     mov        ebp, [esp + 12 + 20]  // YuvConstants
     mov        ecx, [esp + 12 + 24]  // width
@@ skipping 13 matching lines @@
     pop        esi
     ret
   }
 }
 
 __declspec(naked)
 void I422ToRGBARow_SSSE3(const uint8* y_buf,
                          const uint8* u_buf,
                          const uint8* v_buf,
                          uint8* dst_rgba,
+                         struct YuvConstants* yuvconstants,
                          int width) {
   __asm {
     push       esi
     push       edi
-    mov        eax, [esp + 8 + 4]   // Y
-    mov        esi, [esp + 8 + 8]   // U
-    mov        edi, [esp + 8 + 12]  // V
-    mov        edx, [esp + 8 + 16]  // rgba
-    mov        ecx, [esp + 8 + 20]  // width
+    push       ebp
+    mov        eax, [esp + 12 + 4]   // Y
+    mov        esi, [esp + 12 + 8]   // U
+    mov        edi, [esp + 12 + 12]  // V
+    mov        edx, [esp + 12 + 16]  // argb
+    mov        ebp, [esp + 12 + 20]  // YuvConstants
+    mov        ecx, [esp + 12 + 24]  // width
     sub        edi, esi
 
  convertloop:
     READYUV422
-    YUVTORGB(kYuvConstants)
+    YUVTORGB(ebp)
     STORERGBA
 
     sub        ecx, 8
     jg         convertloop
 
+    pop        ebp
     pop        edi
     pop        esi
     ret
   }
 }
-
 #endif  // HAS_I422TOARGBROW_SSSE3
 
 #ifdef HAS_I400TOARGBROW_SSE2
 // 8 pixels of Y converted to 8 pixels of ARGB (32 bytes).
 __declspec(naked)
 void I400ToARGBRow_SSE2(const uint8* y_buf,
                         uint8* rgb_buf,
                         int width) {
   __asm {
     mov        eax, 0x4a354a35      // 4a35 = 18997 = round(1.164 * 64 * 256)
@@ skipping 3382 matching lines @@
 }
 #endif  // HAS_ARGBLUMACOLORTABLEROW_SSSE3
 
 #endif  // defined(_M_X64)
 #endif  // !defined(LIBYUV_DISABLE_X86) && (defined(_M_IX86) || defined(_M_X64))
 
 #ifdef __cplusplus
 }  // extern "C"
 }  // namespace libyuv
 #endif