H264: Fix decoding with CABAC/delta_qp/PCM macroblocks.

[ffmpeg] / libswscale / rgb2rgb.c

diff --git a/libswscale/rgb2rgb.c b/libswscale/rgb2rgb.c
index 11286463292d331ed27e927253d20a1c65f8b47b..14c40700ce6496f307115783f4831678e0253835 100644 (file)
--- a/libswscale/rgb2rgb.c
+++ b/libswscale/rgb2rgb.c
@@ -1,10 +1,10 @@
 /*
- *  rgb2rgb.c, Software RGB to RGB convertor
- *  pluralize by Software PAL8 to RGB convertor
- *               Software YUV to YUV convertor
- *               Software YUV to RGB convertor
- *  Written by Nick Kurshev.
- *  palette & YUV & runtime CPU stuff by Michael (michaelni@gmx.at)
+ * software RGB to RGB converter
+ * pluralize by software PAL8 to RGB converter
+ *              software YUV to YUV converter
+ *              software YUV to RGB converter
+ * Written by Nick Kurshev.
+ * palette & YUV & runtime CPU stuff by Michael (michaelni@gmx.at)
  *
  * This file is part of FFmpeg.
  *
@@ -22,18 +22,18 @@
  * along with FFmpeg; if not, write to the Free Software
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  *
- * the C code (not assembly, mmx, ...) of this file can be used
- * under the LGPL license too
+ * The C code (not assembly, MMX, ...) of this file can be used
+ * under the LGPL license.
  */
 #include <inttypes.h>
 #include "config.h"
+#include "libavutil/x86_cpu.h"
+#include "libavutil/bswap.h"
 #include "rgb2rgb.h"
 #include "swscale.h"
 #include "swscale_internal.h"
-#include "x86_cpu.h"
-#include "bswap.h"
 
-#define FAST_BGR2YV12 // use 7 bit coeffs instead of 15bit
+#define FAST_BGR2YV12 // use 7-bit instead of 15-bit coefficients
 
 void (*rgb24to32)(const uint8_t *src, uint8_t *dst, long src_size);
 void (*rgb24to16)(const uint8_t *src, uint8_t *dst, long src_size);
@@ -122,19 +122,6 @@ DECLARE_ASM_CONST(8, uint64_t, red_15mask)   = 0x00007c0000007c00ULL;
 DECLARE_ASM_CONST(8, uint64_t, green_15mask) = 0x000003e0000003e0ULL;
 DECLARE_ASM_CONST(8, uint64_t, blue_15mask)  = 0x0000001f0000001fULL;
 
-#ifdef FAST_BGR2YV12
-DECLARE_ASM_CONST(8, uint64_t, bgr2YCoeff)   = 0x000000210041000DULL;
-DECLARE_ASM_CONST(8, uint64_t, bgr2UCoeff)   = 0x0000FFEEFFDC0038ULL;
-DECLARE_ASM_CONST(8, uint64_t, bgr2VCoeff)   = 0x00000038FFD2FFF8ULL;
-#else
-DECLARE_ASM_CONST(8, uint64_t, bgr2YCoeff)   = 0x000020E540830C8BULL;
-DECLARE_ASM_CONST(8, uint64_t, bgr2UCoeff)   = 0x0000ED0FDAC23831ULL;
-DECLARE_ASM_CONST(8, uint64_t, bgr2VCoeff)   = 0x00003831D0E6F6EAULL;
-#endif
-DECLARE_ASM_CONST(8, uint64_t, bgr2YOffset)  = 0x1010101010101010ULL;
-DECLARE_ASM_CONST(8, uint64_t, bgr2UVOffset) = 0x8080808080808080ULL;
-DECLARE_ASM_CONST(8, uint64_t, w1111)        = 0x0001000100010001ULL;
-
 #if 0
 static volatile uint64_t __attribute__((aligned(8))) b5Dither;
 static volatile uint64_t __attribute__((aligned(8))) g5Dither;
@@ -162,8 +149,8 @@ static uint64_t __attribute__((aligned(8))) dither8[2]={
 #define RV ((int)( 0.439*(1<<RGB2YUV_SHIFT)+0.5))
 #define RU ((int)(-0.148*(1<<RGB2YUV_SHIFT)+0.5))
 
-//Note: we have C, MMX, MMX2, 3DNOW version therse no 3DNOW+MMX2 one
-//Plain C versions
+//Note: We have C, MMX, MMX2, 3DNOW versions, there is no 3DNOW + MMX2 one.
+//plain C versions
 #undef HAVE_MMX
 #undef HAVE_MMX2
 #undef HAVE_3DNOW
@@ -203,10 +190,10 @@ static uint64_t __attribute__((aligned(8))) dither8[2]={
 #endif //ARCH_X86 || ARCH_X86_64
 
 /*
- rgb15->rgb16 Original by Strepto/Astral
+ RGB15->RGB16 original by Strepto/Astral
  ported to gcc & bugfixed : A'rpi
  MMX2, 3DNOW optimization by Nick Kurshev
- 32bit c version, and and&add trick by Michael Niedermayer
+ 32-bit C version, and and&add trick by Michael Niedermayer
 */
 
 void sws_rgb2rgb_init(int flags){
@@ -279,7 +266,7 @@ void palette8torgb24(const uint8_t *src, uint8_t *dst, long num_pixels, const ui
 {
     long i;
 /*
-    writes 1 byte o much and might cause alignment issues on some architectures?
+    Writes 1 byte too much and might cause alignment issues on some architectures?
     for (i=0; i<num_pixels; i++)
         ((unsigned *)(&dst[i*3])) = ((unsigned *)palette)[src[i]];
 */
@@ -297,7 +284,7 @@ void palette8tobgr24(const uint8_t *src, uint8_t *dst, long num_pixels, const ui
 {
     long i;
 /*
-    writes 1 byte o much and might cause alignment issues on some architectures?
+    Writes 1 byte too much and might cause alignment issues on some architectures?
     for (i=0; i<num_pixels; i++)
         ((unsigned *)(&dst[i*3])) = ((unsigned *)palette)[src[i]];
 */
@@ -312,19 +299,19 @@ void palette8tobgr24(const uint8_t *src, uint8_t *dst, long num_pixels, const ui
 }
 
 /**
- * Palette is assumed to contain bgr16, see rgb32to16 to convert the palette
+ * Palette is assumed to contain BGR16, see rgb32to16 to convert the palette.
  */
 void palette8torgb16(const uint8_t *src, uint8_t *dst, long num_pixels, const uint8_t *palette)
 {
     long i;
     for (i=0; i<num_pixels; i++)
-        ((uint16_t *)dst)[i] = ((uint16_t *)palette)[src[i]];
+        ((uint16_t *)dst)[i] = ((const uint16_t *)palette)[src[i]];
 }
 void palette8tobgr16(const uint8_t *src, uint8_t *dst, long num_pixels, const uint8_t *palette)
 {
     long i;
     for (i=0; i<num_pixels; i++)
-        ((uint16_t *)dst)[i] = bswap_16(((uint16_t *)palette)[src[i]]);
+        ((uint16_t *)dst)[i] = bswap_16(((const uint16_t *)palette)[src[i]]);
 }
 
 /**
@@ -334,13 +321,13 @@ void palette8torgb15(const uint8_t *src, uint8_t *dst, long num_pixels, const ui
 {
     long i;
     for (i=0; i<num_pixels; i++)
-        ((uint16_t *)dst)[i] = ((uint16_t *)palette)[src[i]];
+        ((uint16_t *)dst)[i] = ((const uint16_t *)palette)[src[i]];
 }
 void palette8tobgr15(const uint8_t *src, uint8_t *dst, long num_pixels, const uint8_t *palette)
 {
     long i;
     for (i=0; i<num_pixels; i++)
-        ((uint16_t *)dst)[i] = bswap_16(((uint16_t *)palette)[src[i]]);
+        ((uint16_t *)dst)[i] = bswap_16(((const uint16_t *)palette)[src[i]]);
 }
 
 void rgb32tobgr24(const uint8_t *src, uint8_t *dst, long src_size)
@@ -385,8 +372,8 @@ void rgb24tobgr32(const uint8_t *src, uint8_t *dst, long src_size)
 void rgb16tobgr32(const uint8_t *src, uint8_t *dst, long src_size)
 {
     const uint16_t *end;
-    uint8_t *d = (uint8_t *)dst;
-    const uint16_t *s = (uint16_t *)src;
+    uint8_t *d = dst;
+    const uint16_t *s = (const uint16_t *)src;
     end = s + src_size/2;
     while (s < end)
     {
@@ -409,7 +396,7 @@ void rgb16tobgr32(const uint8_t *src, uint8_t *dst, long src_size)
 void rgb16tobgr24(const uint8_t *src, uint8_t *dst, long src_size)
 {
     const uint16_t *end;
-    uint8_t *d = (uint8_t *)dst;
+    uint8_t *d = dst;
     const uint16_t *s = (const uint16_t *)src;
     end = s + src_size/2;
     while (s < end)
@@ -459,7 +446,7 @@ void rgb16tobgr15(const uint8_t *src, uint8_t *dst, long src_size)
 void rgb15tobgr32(const uint8_t *src, uint8_t *dst, long src_size)
 {
     const uint16_t *end;
-    uint8_t *d = (uint8_t *)dst;
+    uint8_t *d = dst;
     const uint16_t *s = (const uint16_t *)src;
     end = s + src_size/2;
     while (s < end)
@@ -483,8 +470,8 @@ void rgb15tobgr32(const uint8_t *src, uint8_t *dst, long src_size)
 void rgb15tobgr24(const uint8_t *src, uint8_t *dst, long src_size)
 {
     const uint16_t *end;
-    uint8_t *d = (uint8_t *)dst;
-    const uint16_t *s = (uint16_t *)src;
+    uint8_t *d = dst;
+    const uint16_t *s = (const uint16_t *)src;
     end = s + src_size/2;
     while (s < end)
     {