Qt4 - Remove the enqueue button and regroup it under the play button, since it is...
[vlc.git] / modules / video_chroma / i420_rgb16.c
index caa4f92..45f7682 100644 (file)
@@ -1,8 +1,8 @@
 /*****************************************************************************
  * i420_rgb16.c : YUV to bitmap RGB conversion module for vlc
  *****************************************************************************
- * Copyright (C) 2000 VideoLAN
- * $Id: i420_rgb16.c,v 1.4 2002/11/25 19:29:10 sam Exp $
+ * Copyright (C) 2000 the VideoLAN team
+ * $Id$
  *
  * Authors: Samuel Hocevar <sam@zoy.org>
  *
@@ -18,7 +18,7 @@
  *
  * You should have received a copy of the GNU General Public License
  * along with this program; if not, write to the Free Software
- * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA  02111, USA.
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston MA 02110-1301, USA.
  *****************************************************************************/
 
 /*****************************************************************************
 #include <stdlib.h>                                      /* malloc(), free() */
 
 #include <vlc/vlc.h>
-#include <vlc/vout.h>
+#include <vlc_vout.h>
 
 #include "i420_rgb.h"
 #if defined (MODULE_NAME_IS_i420_rgb)
 #   include "i420_rgb_c.h"
 #elif defined (MODULE_NAME_IS_i420_rgb_mmx)
+#   if defined(HAVE_MMX_INTRINSICS)
+#       include <mmintrin.h>
+#   endif
 #   include "i420_rgb_mmx.h"
 #endif
 
-static void SetOffset( int, int, int, int, vlc_bool_t *, int *, int * );
+static void SetOffset( int, int, int, int, vlc_bool_t *,
+                       unsigned int *, int * );
 
 #if defined (MODULE_NAME_IS_i420_rgb)
 /*****************************************************************************
@@ -54,13 +58,13 @@ void E_(I420_RGB16_dithering)( vout_thread_t *p_vout, picture_t *p_src,
                                                       picture_t *p_dest )
 {
     /* We got this one from the old arguments */
-    u16 *p_pic = (u16*)p_dest->p->p_pixels;
-    u8  *p_y   = p_src->Y_PIXELS;
-    u8  *p_u   = p_src->U_PIXELS;
-    u8  *p_v   = p_src->V_PIXELS;
+    uint16_t *p_pic = (uint16_t*)p_dest->p->p_pixels;
+    uint8_t  *p_y   = p_src->Y_PIXELS;
+    uint8_t  *p_u   = p_src->U_PIXELS;
+    uint8_t  *p_v   = p_src->V_PIXELS;
 
     vlc_bool_t   b_hscale;                        /* horizontal scaling type */
-    int          i_vscale;                          /* vertical scaling type */
+    unsigned int i_vscale;                          /* vertical scaling type */
     unsigned int i_x, i_y;                /* horizontal and vertical indexes */
     unsigned int i_real_y;                                          /* y % 4 */
 
@@ -68,20 +72,25 @@ void E_(I420_RGB16_dithering)( vout_thread_t *p_vout, picture_t *p_src,
     int         i_rewind;
     int         i_scale_count;                       /* scale modulo counter */
     int         i_chroma_width = p_vout->render.i_width / 2; /* chroma width */
-    u16 *       p_pic_start;       /* beginning of the current line for copy */
+    uint16_t *  p_pic_start;       /* beginning of the current line for copy */
     int         i_uval, i_vval;                           /* U and V samples */
     int         i_red, i_green, i_blue;          /* U and V modified samples */
-    u16 *       p_yuv = p_vout->chroma.p_sys->p_rgb16;
-    u16 *       p_ybase;                     /* Y dependant conversion table */
+    uint16_t *  p_yuv = p_vout->chroma.p_sys->p_rgb16;
+    uint16_t *  p_ybase;                     /* Y dependant conversion table */
 
     /* Conversion buffer pointer */
-    u16 *       p_buffer_start = (u16*)p_vout->chroma.p_sys->p_buffer;
-    u16 *       p_buffer;
+    uint16_t *  p_buffer_start = (uint16_t*)p_vout->chroma.p_sys->p_buffer;
+    uint16_t *  p_buffer;
 
     /* Offset array pointer */
     int *       p_offset_start = p_vout->chroma.p_sys->p_offset;
     int *       p_offset;
 
+    const int i_source_margin = p_src->p[0].i_pitch
+                                 - p_src->p[0].i_visible_pitch;
+    const int i_source_margin_c = p_src->p[1].i_pitch
+                                 - p_src->p[1].i_visible_pitch;
+
     /* The dithering matrices */
     int dither10[4] = {  0x0,  0x8,  0x2,  0xa };
     int dither11[4] = {  0xc,  0x4,  0xe,  0x6 };
@@ -172,6 +181,13 @@ void E_(I420_RGB16_dithering)( vout_thread_t *p_vout, picture_t *p_src,
         }
         SCALE_WIDTH;
         SCALE_HEIGHT( 420, 2 );
+
+        p_y += i_source_margin;
+        if( i_y % 2 )
+        {
+            p_u += i_source_margin_c;
+            p_v += i_source_margin_c;
+        }
     }
 }
 #endif
@@ -190,10 +206,10 @@ void E_(I420_RGB16)( vout_thread_t *p_vout, picture_t *p_src,
                                             picture_t *p_dest )
 {
     /* We got this one from the old arguments */
-    u16 *p_pic = (u16*)p_dest->p->p_pixels;
-    u8  *p_y   = p_src->Y_PIXELS;
-    u8  *p_u   = p_src->U_PIXELS;
-    u8  *p_v   = p_src->V_PIXELS;
+    uint16_t *p_pic = (uint16_t*)p_dest->p->p_pixels;
+    uint8_t  *p_y   = p_src->Y_PIXELS;
+    uint8_t  *p_u   = p_src->U_PIXELS;
+    uint8_t  *p_v   = p_src->V_PIXELS;
 
     vlc_bool_t  b_hscale;                         /* horizontal scaling type */
     unsigned int i_vscale;                          /* vertical scaling type */
@@ -203,22 +219,27 @@ void E_(I420_RGB16)( vout_thread_t *p_vout, picture_t *p_src,
     int         i_rewind;
     int         i_scale_count;                       /* scale modulo counter */
     int         i_chroma_width = p_vout->render.i_width / 2; /* chroma width */
-    u16 *       p_pic_start;       /* beginning of the current line for copy */
+    uint16_t *  p_pic_start;       /* beginning of the current line for copy */
 #if defined (MODULE_NAME_IS_i420_rgb)
     int         i_uval, i_vval;                           /* U and V samples */
     int         i_red, i_green, i_blue;          /* U and V modified samples */
-    u16 *       p_yuv = p_vout->chroma.p_sys->p_rgb16;
-    u16 *       p_ybase;                     /* Y dependant conversion table */
+    uint16_t *  p_yuv = p_vout->chroma.p_sys->p_rgb16;
+    uint16_t *  p_ybase;                     /* Y dependant conversion table */
 #endif
 
     /* Conversion buffer pointer */
-    u16 *       p_buffer_start = (u16*)p_vout->chroma.p_sys->p_buffer;
-    u16 *       p_buffer;
+    uint16_t *  p_buffer_start = (uint16_t*)p_vout->chroma.p_sys->p_buffer;
+    uint16_t *  p_buffer;
 
     /* Offset array pointer */
     int *       p_offset_start = p_vout->chroma.p_sys->p_offset;
     int *       p_offset;
 
+    const int i_source_margin = p_src->p[0].i_pitch
+                                 - p_src->p[0].i_visible_pitch;
+    const int i_source_margin_c = p_src->p[1].i_pitch
+                                 - p_src->p[1].i_visible_pitch;
+
     i_right_margin = p_dest->p->i_pitch - p_dest->p->i_visible_pitch;
 
     if( p_vout->render.i_width & 7 )
@@ -261,14 +282,23 @@ void E_(I420_RGB16)( vout_thread_t *p_vout, picture_t *p_src,
             /* 15bpp 5/5/5 */
             for ( i_x = p_vout->render.i_width / 8; i_x--; )
             {
+#   if defined (HAVE_MMX_INTRINSICS)
+                __m64 mm0, mm1, mm2, mm3, mm4, mm5, mm6, mm7;
+                uint64_t tmp64;
+                INTRINSICS_INIT_16
+                INTRINSICS_YUV_MUL
+                INTRINSICS_YUV_ADD
+                INTRINSICS_UNPACK_15
+#   else
                 __asm__( MMX_INIT_16
                          : : "r" (p_y), "r" (p_u), "r" (p_v), "r" (p_buffer) );
 
-                __asm__( ".align 8"
+                __asm__( ".p2align 3"
                          MMX_YUV_MUL
                          MMX_YUV_ADD
                          MMX_UNPACK_15
                          : : "r" (p_y), "r" (p_u), "r" (p_v), "r" (p_buffer) );
+#   endif
 
                 p_y += 8;
                 p_u += 4;
@@ -281,14 +311,23 @@ void E_(I420_RGB16)( vout_thread_t *p_vout, picture_t *p_src,
             /* 16bpp 5/6/5 */
             for ( i_x = p_vout->render.i_width / 8; i_x--; )
             {
+#   if defined (HAVE_MMX_INTRINSICS)
+                __m64 mm0, mm1, mm2, mm3, mm4, mm5, mm6, mm7;
+                uint64_t tmp64;
+                INTRINSICS_INIT_16
+                INTRINSICS_YUV_MUL
+                INTRINSICS_YUV_ADD
+                INTRINSICS_UNPACK_16
+#   else
                 __asm__( MMX_INIT_16
                          : : "r" (p_y), "r" (p_u), "r" (p_v), "r" (p_buffer) );
 
-                __asm__( ".align 8"
+                __asm__( ".p2align 3"
                          MMX_YUV_MUL
                          MMX_YUV_ADD
                          MMX_UNPACK_16
                          : : "r" (p_y), "r" (p_u), "r" (p_v), "r" (p_buffer) );
+#   endif
 
                 p_y += 8;
                 p_u += 4;
@@ -302,6 +341,12 @@ void E_(I420_RGB16)( vout_thread_t *p_vout, picture_t *p_src,
          * at least we have all the pixels */
         if( i_rewind )
         {
+#if defined (MODULE_NAME_IS_i420_rgb_mmx)
+#   if defined (HAVE_MMX_INTRINSICS)
+            __m64 mm0, mm1, mm2, mm3, mm4, mm5, mm6, mm7;
+            uint64_t tmp64;
+#   endif
+#endif
             p_y -= i_rewind;
             p_u -= i_rewind >> 1;
             p_v -= i_rewind >> 1;
@@ -312,26 +357,43 @@ void E_(I420_RGB16)( vout_thread_t *p_vout, picture_t *p_src,
             CONVERT_YUV_PIXEL(2);  CONVERT_Y_PIXEL(2);
             CONVERT_YUV_PIXEL(2);  CONVERT_Y_PIXEL(2);
 #elif defined (MODULE_NAME_IS_i420_rgb_mmx)
+
+#   if defined (HAVE_MMX_INTRINSICS)
+            INTRINSICS_INIT_16
+#   else
             __asm__( MMX_INIT_16
                      : : "r" (p_y), "r" (p_u), "r" (p_v), "r" (p_buffer) );
+#   endif
 
             if( p_vout->output.i_rmask == 0x7c00 )
             {
                 /* 15bpp 5/5/5 */
-                __asm__( ".align 8"
+#   if defined (HAVE_MMX_INTRINSICS)
+                INTRINSICS_YUV_MUL
+                INTRINSICS_YUV_ADD
+                INTRINSICS_UNPACK_15
+#   else
+                __asm__( ".p2align 3"
                          MMX_YUV_MUL
                          MMX_YUV_ADD
                          MMX_UNPACK_15
                          : : "r" (p_y), "r" (p_u), "r" (p_v), "r" (p_buffer) );
+#   endif
             }
             else
             {
+#   if defined (HAVE_MMX_INTRINSICS)
+                INTRINSICS_YUV_MUL
+                INTRINSICS_YUV_ADD
+                INTRINSICS_UNPACK_16
+#   else
                 /* 16bpp 5/6/5 */
-                __asm__( ".align 8"
+                __asm__( ".p2align 3"
                          MMX_YUV_MUL
                          MMX_YUV_ADD
                          MMX_UNPACK_16
                          : : "r" (p_y), "r" (p_u), "r" (p_v), "r" (p_buffer) );
+#   endif
             }
 
             p_y += 8;
@@ -342,6 +404,13 @@ void E_(I420_RGB16)( vout_thread_t *p_vout, picture_t *p_src,
         }
         SCALE_WIDTH;
         SCALE_HEIGHT( 420, 2 );
+
+        p_y += i_source_margin;
+        if( i_y % 2 )
+        {
+            p_u += i_source_margin_c;
+            p_v += i_source_margin_c;
+        }
     }
 }
 
@@ -359,10 +428,10 @@ void E_(I420_RGB32)( vout_thread_t *p_vout, picture_t *p_src,
                                             picture_t *p_dest )
 {
     /* We got this one from the old arguments */
-    u32 *p_pic = (u32*)p_dest->p->p_pixels;
-    u8  *p_y   = p_src->Y_PIXELS;
-    u8  *p_u   = p_src->U_PIXELS;
-    u8  *p_v   = p_src->V_PIXELS;
+    uint32_t *p_pic = (uint32_t*)p_dest->p->p_pixels;
+    uint8_t  *p_y   = p_src->Y_PIXELS;
+    uint8_t  *p_u   = p_src->U_PIXELS;
+    uint8_t  *p_v   = p_src->V_PIXELS;
 
     vlc_bool_t  b_hscale;                         /* horizontal scaling type */
     unsigned int i_vscale;                          /* vertical scaling type */
@@ -372,22 +441,27 @@ void E_(I420_RGB32)( vout_thread_t *p_vout, picture_t *p_src,
     int         i_rewind;
     int         i_scale_count;                       /* scale modulo counter */
     int         i_chroma_width = p_vout->render.i_width / 2; /* chroma width */
-    u32 *       p_pic_start;       /* beginning of the current line for copy */
+    uint32_t *  p_pic_start;       /* beginning of the current line for copy */
 #if defined (MODULE_NAME_IS_i420_rgb)
     int         i_uval, i_vval;                           /* U and V samples */
     int         i_red, i_green, i_blue;          /* U and V modified samples */
-    u32 *       p_yuv = p_vout->chroma.p_sys->p_rgb32;
-    u32 *       p_ybase;                     /* Y dependant conversion table */
+    uint32_t *  p_yuv = p_vout->chroma.p_sys->p_rgb32;
+    uint32_t *  p_ybase;                     /* Y dependant conversion table */
 #endif
 
     /* Conversion buffer pointer */
-    u32 *       p_buffer_start = (u32*)p_vout->chroma.p_sys->p_buffer;
-    u32 *       p_buffer;
+    uint32_t *  p_buffer_start = (uint32_t*)p_vout->chroma.p_sys->p_buffer;
+    uint32_t *  p_buffer;
 
     /* Offset array pointer */
     int *       p_offset_start = p_vout->chroma.p_sys->p_offset;
     int *       p_offset;
 
+    const int i_source_margin = p_src->p[0].i_pitch
+                                 - p_src->p[0].i_visible_pitch;
+    const int i_source_margin_c = p_src->p[1].i_pitch
+                                 - p_src->p[1].i_visible_pitch;
+
     i_right_margin = p_dest->p->i_pitch - p_dest->p->i_visible_pitch;
 
     if( p_vout->render.i_width & 7 )
@@ -424,14 +498,23 @@ void E_(I420_RGB32)( vout_thread_t *p_vout, picture_t *p_src,
             CONVERT_YUV_PIXEL(4);  CONVERT_Y_PIXEL(4);
             CONVERT_YUV_PIXEL(4);  CONVERT_Y_PIXEL(4);
 #elif defined (MODULE_NAME_IS_i420_rgb_mmx)
+#   if defined (HAVE_MMX_INTRINSICS)
+            __m64 mm0, mm1, mm2, mm3, mm4, mm5, mm6, mm7;
+            uint64_t tmp64;
+            INTRINSICS_INIT_32
+            INTRINSICS_YUV_MUL
+            INTRINSICS_YUV_ADD
+            INTRINSICS_UNPACK_32
+#   else
             __asm__( MMX_INIT_32
                      : : "r" (p_y), "r" (p_u), "r" (p_v), "r" (p_buffer) );
 
-            __asm__( ".align 8"
+            __asm__( ".p2align 3"
                      MMX_YUV_MUL
                      MMX_YUV_ADD
                      MMX_UNPACK_32
                      : : "r" (p_y), "r" (p_u), "r" (p_v), "r" (p_buffer) );
+#   endif
 
             p_y += 8;
             p_u += 4;
@@ -444,6 +527,12 @@ void E_(I420_RGB32)( vout_thread_t *p_vout, picture_t *p_src,
          * at least we have all the pixels */
         if( i_rewind )
         {
+#if defined (MODULE_NAME_IS_i420_rgb_mmx)
+#   if defined (HAVE_MMX_INTRINSICS)
+            __m64 mm0, mm1, mm2, mm3, mm4, mm5, mm6, mm7;
+            uint64_t tmp64;
+#   endif
+#endif
             p_y -= i_rewind;
             p_u -= i_rewind >> 1;
             p_v -= i_rewind >> 1;
@@ -454,14 +543,21 @@ void E_(I420_RGB32)( vout_thread_t *p_vout, picture_t *p_src,
             CONVERT_YUV_PIXEL(4);  CONVERT_Y_PIXEL(4);
             CONVERT_YUV_PIXEL(4);  CONVERT_Y_PIXEL(4);
 #elif defined (MODULE_NAME_IS_i420_rgb_mmx)
+#   if defined (HAVE_MMX_INTRINSICS)
+            INTRINSICS_INIT_32
+            INTRINSICS_YUV_MUL
+            INTRINSICS_YUV_ADD
+            INTRINSICS_UNPACK_32
+#   else
             __asm__( MMX_INIT_32
                      : : "r" (p_y), "r" (p_u), "r" (p_v), "r" (p_buffer) );
 
-            __asm__( ".align 8"
+            __asm__( ".p2align 3"
                      MMX_YUV_MUL
                      MMX_YUV_ADD
                      MMX_UNPACK_32
                      : : "r" (p_y), "r" (p_u), "r" (p_v), "r" (p_buffer) );
+#   endif
 
             p_y += 8;
             p_u += 4;
@@ -471,6 +567,13 @@ void E_(I420_RGB32)( vout_thread_t *p_vout, picture_t *p_src,
         }
         SCALE_WIDTH;
         SCALE_HEIGHT( 420, 4 );
+
+        p_y += i_source_margin;
+        if( i_y % 2 )
+        {
+            p_u += i_source_margin_c;
+            p_v += i_source_margin_c;
+        }
     }
 }
 
@@ -484,7 +587,7 @@ void E_(I420_RGB32)( vout_thread_t *p_vout, picture_t *p_src,
  *****************************************************************************/
 static void SetOffset( int i_width, int i_height, int i_pic_width,
                        int i_pic_height, vlc_bool_t *pb_hscale,
-                       int *pi_vscale, int *p_offset )
+                       unsigned int *pi_vscale, int *p_offset )
 {
     int i_x;                                    /* x position in destination */
     int i_scale_count;                                     /* modulo counter */