/*
- * 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.
*
*
* You should have received a copy of the GNU General Public License
* along with FFmpeg; if not, write to the Free Software
- * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
+ * 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
-
-void (*rgb24to32)(const uint8_t *src,uint8_t *dst,long src_size);
-void (*rgb24to16)(const uint8_t *src,uint8_t *dst,long src_size);
-void (*rgb24to15)(const uint8_t *src,uint8_t *dst,long src_size);
-void (*rgb32to24)(const uint8_t *src,uint8_t *dst,long src_size);
-void (*rgb32to16)(const uint8_t *src,uint8_t *dst,long src_size);
-void (*rgb32to15)(const uint8_t *src,uint8_t *dst,long src_size);
-void (*rgb15to16)(const uint8_t *src,uint8_t *dst,long src_size);
-void (*rgb15to24)(const uint8_t *src,uint8_t *dst,long src_size);
-void (*rgb15to32)(const uint8_t *src,uint8_t *dst,long src_size);
-void (*rgb16to15)(const uint8_t *src,uint8_t *dst,long src_size);
-void (*rgb16to24)(const uint8_t *src,uint8_t *dst,long src_size);
-void (*rgb16to32)(const uint8_t *src,uint8_t *dst,long src_size);
+
+#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);
+void (*rgb24to15)(const uint8_t *src, uint8_t *dst, long src_size);
+void (*rgb32to24)(const uint8_t *src, uint8_t *dst, long src_size);
+void (*rgb32to16)(const uint8_t *src, uint8_t *dst, long src_size);
+void (*rgb32to15)(const uint8_t *src, uint8_t *dst, long src_size);
+void (*rgb15to16)(const uint8_t *src, uint8_t *dst, long src_size);
+void (*rgb15to24)(const uint8_t *src, uint8_t *dst, long src_size);
+void (*rgb15to32)(const uint8_t *src, uint8_t *dst, long src_size);
+void (*rgb16to15)(const uint8_t *src, uint8_t *dst, long src_size);
+void (*rgb16to24)(const uint8_t *src, uint8_t *dst, long src_size);
+void (*rgb16to32)(const uint8_t *src, uint8_t *dst, long src_size);
//void (*rgb24tobgr32)(const uint8_t *src, uint8_t *dst, long src_size);
void (*rgb24tobgr24)(const uint8_t *src, uint8_t *dst, long src_size);
void (*rgb24tobgr16)(const uint8_t *src, uint8_t *dst, long src_size);
long srcStride3, long dstStride);
#if defined(ARCH_X86) && defined(CONFIG_GPL)
-static const uint64_t mmx_null __attribute__((aligned(8))) = 0x0000000000000000ULL;
-static const uint64_t mmx_one __attribute__((aligned(8))) = 0xFFFFFFFFFFFFFFFFULL;
-static const uint64_t mask32b attribute_used __attribute__((aligned(8))) = 0x000000FF000000FFULL;
-static const uint64_t mask32g attribute_used __attribute__((aligned(8))) = 0x0000FF000000FF00ULL;
-static const uint64_t mask32r attribute_used __attribute__((aligned(8))) = 0x00FF000000FF0000ULL;
-static const uint64_t mask32 __attribute__((aligned(8))) = 0x00FFFFFF00FFFFFFULL;
-static const uint64_t mask3216br __attribute__((aligned(8))) = 0x00F800F800F800F8ULL;
-static const uint64_t mask3216g __attribute__((aligned(8))) = 0x0000FC000000FC00ULL;
-static const uint64_t mask3215g __attribute__((aligned(8))) = 0x0000F8000000F800ULL;
-static const uint64_t mul3216 __attribute__((aligned(8))) = 0x2000000420000004ULL;
-static const uint64_t mul3215 __attribute__((aligned(8))) = 0x2000000820000008ULL;
-static const uint64_t mask24b attribute_used __attribute__((aligned(8))) = 0x00FF0000FF0000FFULL;
-static const uint64_t mask24g attribute_used __attribute__((aligned(8))) = 0xFF0000FF0000FF00ULL;
-static const uint64_t mask24r attribute_used __attribute__((aligned(8))) = 0x0000FF0000FF0000ULL;
-static const uint64_t mask24l __attribute__((aligned(8))) = 0x0000000000FFFFFFULL;
-static const uint64_t mask24h __attribute__((aligned(8))) = 0x0000FFFFFF000000ULL;
-static const uint64_t mask24hh __attribute__((aligned(8))) = 0xffff000000000000ULL;
-static const uint64_t mask24hhh __attribute__((aligned(8))) = 0xffffffff00000000ULL;
-static const uint64_t mask24hhhh __attribute__((aligned(8))) = 0xffffffffffff0000ULL;
-static const uint64_t mask15b __attribute__((aligned(8))) = 0x001F001F001F001FULL; /* 00000000 00011111 xxB */
-static const uint64_t mask15rg __attribute__((aligned(8))) = 0x7FE07FE07FE07FE0ULL; /* 01111111 11100000 RGx */
-static const uint64_t mask15s __attribute__((aligned(8))) = 0xFFE0FFE0FFE0FFE0ULL;
-static const uint64_t mask15g __attribute__((aligned(8))) = 0x03E003E003E003E0ULL;
-static const uint64_t mask15r __attribute__((aligned(8))) = 0x7C007C007C007C00ULL;
+DECLARE_ASM_CONST(8, uint64_t, mmx_null) = 0x0000000000000000ULL;
+DECLARE_ASM_CONST(8, uint64_t, mmx_one) = 0xFFFFFFFFFFFFFFFFULL;
+DECLARE_ASM_CONST(8, uint64_t, mask32b) = 0x000000FF000000FFULL;
+DECLARE_ASM_CONST(8, uint64_t, mask32g) = 0x0000FF000000FF00ULL;
+DECLARE_ASM_CONST(8, uint64_t, mask32r) = 0x00FF000000FF0000ULL;
+DECLARE_ASM_CONST(8, uint64_t, mask32) = 0x00FFFFFF00FFFFFFULL;
+DECLARE_ASM_CONST(8, uint64_t, mask3216br) = 0x00F800F800F800F8ULL;
+DECLARE_ASM_CONST(8, uint64_t, mask3216g) = 0x0000FC000000FC00ULL;
+DECLARE_ASM_CONST(8, uint64_t, mask3215g) = 0x0000F8000000F800ULL;
+DECLARE_ASM_CONST(8, uint64_t, mul3216) = 0x2000000420000004ULL;
+DECLARE_ASM_CONST(8, uint64_t, mul3215) = 0x2000000820000008ULL;
+DECLARE_ASM_CONST(8, uint64_t, mask24b) = 0x00FF0000FF0000FFULL;
+DECLARE_ASM_CONST(8, uint64_t, mask24g) = 0xFF0000FF0000FF00ULL;
+DECLARE_ASM_CONST(8, uint64_t, mask24r) = 0x0000FF0000FF0000ULL;
+DECLARE_ASM_CONST(8, uint64_t, mask24l) = 0x0000000000FFFFFFULL;
+DECLARE_ASM_CONST(8, uint64_t, mask24h) = 0x0000FFFFFF000000ULL;
+DECLARE_ASM_CONST(8, uint64_t, mask24hh) = 0xffff000000000000ULL;
+DECLARE_ASM_CONST(8, uint64_t, mask24hhh) = 0xffffffff00000000ULL;
+DECLARE_ASM_CONST(8, uint64_t, mask24hhhh) = 0xffffffffffff0000ULL;
+DECLARE_ASM_CONST(8, uint64_t, mask15b) = 0x001F001F001F001FULL; /* 00000000 00011111 xxB */
+DECLARE_ASM_CONST(8, uint64_t, mask15rg) = 0x7FE07FE07FE07FE0ULL; /* 01111111 11100000 RGx */
+DECLARE_ASM_CONST(8, uint64_t, mask15s) = 0xFFE0FFE0FFE0FFE0ULL;
+DECLARE_ASM_CONST(8, uint64_t, mask15g) = 0x03E003E003E003E0ULL;
+DECLARE_ASM_CONST(8, uint64_t, mask15r) = 0x7C007C007C007C00ULL;
#define mask16b mask15b
-static const uint64_t mask16g __attribute__((aligned(8))) = 0x07E007E007E007E0ULL;
-static const uint64_t mask16r __attribute__((aligned(8))) = 0xF800F800F800F800ULL;
-static const uint64_t red_16mask __attribute__((aligned(8))) = 0x0000f8000000f800ULL;
-static const uint64_t green_16mask __attribute__((aligned(8))) = 0x000007e0000007e0ULL;
-static const uint64_t blue_16mask __attribute__((aligned(8))) = 0x0000001f0000001fULL;
-static const uint64_t red_15mask __attribute__((aligned(8))) = 0x00007c0000007c00ULL;
-static const uint64_t green_15mask __attribute__((aligned(8))) = 0x000003e0000003e0ULL;
-static const uint64_t blue_15mask __attribute__((aligned(8))) = 0x0000001f0000001fULL;
-
-#ifdef FAST_BGR2YV12
-static const uint64_t bgr2YCoeff attribute_used __attribute__((aligned(8))) = 0x000000210041000DULL;
-static const uint64_t bgr2UCoeff attribute_used __attribute__((aligned(8))) = 0x0000FFEEFFDC0038ULL;
-static const uint64_t bgr2VCoeff attribute_used __attribute__((aligned(8))) = 0x00000038FFD2FFF8ULL;
-#else
-static const uint64_t bgr2YCoeff attribute_used __attribute__((aligned(8))) = 0x000020E540830C8BULL;
-static const uint64_t bgr2UCoeff attribute_used __attribute__((aligned(8))) = 0x0000ED0FDAC23831ULL;
-static const uint64_t bgr2VCoeff attribute_used __attribute__((aligned(8))) = 0x00003831D0E6F6EAULL;
-#endif
-static const uint64_t bgr2YOffset attribute_used __attribute__((aligned(8))) = 0x1010101010101010ULL;
-static const uint64_t bgr2UVOffset attribute_used __attribute__((aligned(8))) = 0x8080808080808080ULL;
-static const uint64_t w1111 attribute_used __attribute__((aligned(8))) = 0x0001000100010001ULL;
+DECLARE_ASM_CONST(8, uint64_t, mask16g) = 0x07E007E007E007E0ULL;
+DECLARE_ASM_CONST(8, uint64_t, mask16r) = 0xF800F800F800F800ULL;
+DECLARE_ASM_CONST(8, uint64_t, red_16mask) = 0x0000f8000000f800ULL;
+DECLARE_ASM_CONST(8, uint64_t, green_16mask) = 0x000007e0000007e0ULL;
+DECLARE_ASM_CONST(8, uint64_t, blue_16mask) = 0x0000001f0000001fULL;
+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;
#if 0
static volatile uint64_t __attribute__((aligned(8))) b5Dither;
#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
#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){
/*
for (i=0; i<num_pixels; i++)
- ((unsigned *)dst)[i] = ((unsigned *)palette)[ src[i] ];
+ ((unsigned *)dst)[i] = ((unsigned *)palette)[src[i]];
*/
for (i=0; i<num_pixels; i++)
{
#ifdef WORDS_BIGENDIAN
- dst[3]= palette[ src[i]*4+2 ];
- dst[2]= palette[ src[i]*4+1 ];
- dst[1]= palette[ src[i]*4+0 ];
+ dst[3]= palette[src[i]*4+2];
+ dst[2]= palette[src[i]*4+1];
+ dst[1]= palette[src[i]*4+0];
#else
//FIXME slow?
- dst[0]= palette[ src[i]*4+2 ];
- dst[1]= palette[ src[i]*4+1 ];
- dst[2]= palette[ src[i]*4+0 ];
+ dst[0]= palette[src[i]*4+2];
+ dst[1]= palette[src[i]*4+1];
+ dst[2]= palette[src[i]*4+0];
//dst[3]= 0; /* do we need this cleansing? */
#endif
dst+= 4;
for (i=0; i<num_pixels; i++)
{
#ifdef WORDS_BIGENDIAN
- dst[3]= palette[ src[i]*4+0 ];
- dst[2]= palette[ src[i]*4+1 ];
- dst[1]= palette[ src[i]*4+2 ];
+ dst[3]= palette[src[i]*4+0];
+ dst[2]= palette[src[i]*4+1];
+ dst[1]= palette[src[i]*4+2];
#else
//FIXME slow?
- dst[0]= palette[ src[i]*4+0 ];
- dst[1]= palette[ src[i]*4+1 ];
- dst[2]= palette[ src[i]*4+2 ];
+ dst[0]= palette[src[i]*4+0];
+ dst[1]= palette[src[i]*4+1];
+ dst[2]= palette[src[i]*4+2];
//dst[3]= 0; /* do we need this cleansing? */
#endif
{
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] ];
+ ((unsigned *)(&dst[i*3])) = ((unsigned *)palette)[src[i]];
*/
for (i=0; i<num_pixels; i++)
{
//FIXME slow?
- dst[0]= palette[ src[i]*4+2 ];
- dst[1]= palette[ src[i]*4+1 ];
- dst[2]= palette[ src[i]*4+0 ];
+ dst[0]= palette[src[i]*4+2];
+ dst[1]= palette[src[i]*4+1];
+ dst[2]= palette[src[i]*4+0];
dst+= 3;
}
}
{
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] ];
+ ((unsigned *)(&dst[i*3])) = ((unsigned *)palette)[src[i]];
*/
for (i=0; i<num_pixels; i++)
{
//FIXME slow?
- dst[0]= palette[ src[i]*4+0 ];
- dst[1]= palette[ src[i]*4+1 ];
- dst[2]= palette[ src[i]*4+2 ];
+ dst[0]= palette[src[i]*4+0];
+ dst[1]= palette[src[i]*4+1];
+ dst[2]= palette[src[i]*4+2];
dst+= 3;
}
}
/**
- * 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]]);
}
/**
{
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)
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)
{
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)
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)
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)
{
projects / ffmpeg / blobdiff