ffmpeg roll of source to mar 9 version...

patched-ffmpeg-mt/libavcodec/dsputil.h

 /*
  * DSP utils
  * Copyright (c) 2000, 2001, 2002 Fabrice Bellard
  * Copyright (c) 2002-2004 Michael Niedermayer <michaelni@gmx.at>
  *
  * This file is part of FFmpeg.
  *
  * FFmpeg is free software; you can redistribute it and/or
  * modify it under the terms of the GNU Lesser General Public
  * License as published by the Free Software Foundation; either
@@ skipping 78 matching lines @@
 void ff_vp3_idct_put_c(uint8_t *dest/*align 8*/, int line_size, DCTELEM *block/*align 16*/);
 void ff_vp3_idct_add_c(uint8_t *dest/*align 8*/, int line_size, DCTELEM *block/*align 16*/);
 
 void ff_vp3_v_loop_filter_c(uint8_t *src, int stride, int *bounding_values);
 void ff_vp3_h_loop_filter_c(uint8_t *src, int stride, int *bounding_values);
 
 /* VP6 DSP functions */
 void ff_vp6_filter_diag4_c(uint8_t *dst, uint8_t *src, int stride,
                            const int16_t *h_weights, const int16_t *v_weights);
 
+/* Bink functions */
+void ff_bink_idct_c    (DCTELEM *block);
+void ff_bink_idct_add_c(uint8_t *dest, int linesize, DCTELEM *block);
+void ff_bink_idct_put_c(uint8_t *dest, int linesize, DCTELEM *block);
+
+/* CAVS functions */
+void ff_put_cavs_qpel8_mc00_c(uint8_t *dst, uint8_t *src, int stride);
+void ff_avg_cavs_qpel8_mc00_c(uint8_t *dst, uint8_t *src, int stride);
+void ff_put_cavs_qpel16_mc00_c(uint8_t *dst, uint8_t *src, int stride);
+void ff_avg_cavs_qpel16_mc00_c(uint8_t *dst, uint8_t *src, int stride);
+
+/* VC1 functions */
+void ff_put_vc1_mspel_mc00_c(uint8_t *dst, const uint8_t *src, int stride, int rnd);
+void ff_avg_vc1_mspel_mc00_c(uint8_t *dst, const uint8_t *src, int stride, int rnd);
+
+/* EA functions */
+void ff_ea_idct_put_c(uint8_t *dest, int linesize, DCTELEM *block);
+
 /* 1/2^n downscaling functions from imgconvert.c */
 void ff_img_copy_plane(uint8_t *dst, int dst_wrap, const uint8_t *src, int src_wrap, int width, int height);
 void ff_shrink22(uint8_t *dst, int dst_wrap, const uint8_t *src, int src_wrap, int width, int height);
 void ff_shrink44(uint8_t *dst, int dst_wrap, const uint8_t *src, int src_wrap, int width, int height);
 void ff_shrink88(uint8_t *dst, int dst_wrap, const uint8_t *src, int src_wrap, int width, int height);
 
 void ff_gmc_c(uint8_t *dst, uint8_t *src, int stride, int h, int ox, int oy,
               int dxx, int dxy, int dyx, int dyy, int shift, int r, int width, int height);
 
 /* minimum alignment rules ;)
@@ skipping 20 matching lines @@
 /* add and put pixel (decoding) */
 // blocksizes for op_pixels_func are 8x4,8x8 16x8 16x16
 //h for op_pixels_func is limited to {width/2, width} but never larger than 16 and never smaller then 4
 typedef void (*op_pixels_func)(uint8_t *block/*align width (8 or 16)*/, const uint8_t *pixels/*align 1*/, int line_size, int h);
 typedef void (*tpel_mc_func)(uint8_t *block/*align width (8 or 16)*/, const uint8_t *pixels/*align 1*/, int line_size, int w, int h);
 typedef void (*qpel_mc_func)(uint8_t *dst/*align width (8 or 16)*/, uint8_t *src/*align 1*/, int stride);
 typedef void (*h264_chroma_mc_func)(uint8_t *dst/*align 8*/, uint8_t *src/*align 1*/, int srcStride, int h, int x, int y);
 typedef void (*h264_weight_func)(uint8_t *block, int stride, int log2_denom, int weight, int offset);
 typedef void (*h264_biweight_func)(uint8_t *dst, uint8_t *src, int stride, int log2_denom, int weightd, int weights, int offset);
 
+typedef void (*op_fill_func)(uint8_t *block/*align width (8 or 16)*/, uint8_t value, int line_size, int h);
+
 #define DEF_OLD_QPEL(name)\
 void ff_put_        ## name (uint8_t *dst/*align width (8 or 16)*/, uint8_t *src/*align 1*/, int stride);\
 void ff_put_no_rnd_ ## name (uint8_t *dst/*align width (8 or 16)*/, uint8_t *src/*align 1*/, int stride);\
 void ff_avg_        ## name (uint8_t *dst/*align width (8 or 16)*/, uint8_t *src/*align 1*/, int stride);
 
 DEF_OLD_QPEL(qpel16_mc11_old_c)
 DEF_OLD_QPEL(qpel16_mc31_old_c)
 DEF_OLD_QPEL(qpel16_mc12_old_c)
 DEF_OLD_QPEL(qpel16_mc32_old_c)
 DEF_OLD_QPEL(qpel16_mc13_old_c)
@@ skipping 41 matching lines @@
 
 /**
  * DSPContext.
  */
 typedef struct DSPContext {
     /* pixel ops : interface with DCT */
     void (*get_pixels)(DCTELEM *block/*align 16*/, const uint8_t *pixels/*align 8*/, int line_size);
     void (*diff_pixels)(DCTELEM *block/*align 16*/, const uint8_t *s1/*align 8*/, const uint8_t *s2/*align 8*/, int stride);
     void (*put_pixels_clamped)(const DCTELEM *block/*align 16*/, uint8_t *pixels/*align 8*/, int line_size);
     void (*put_signed_pixels_clamped)(const DCTELEM *block/*align 16*/, uint8_t *pixels/*align 8*/, int line_size);
+    void (*put_pixels_nonclamped)(const DCTELEM *block/*align 16*/, uint8_t *pixels/*align 8*/, int line_size);
     void (*add_pixels_clamped)(const DCTELEM *block/*align 16*/, uint8_t *pixels/*align 8*/, int line_size);
     void (*add_pixels8)(uint8_t *pixels, DCTELEM *block, int line_size);
     void (*add_pixels4)(uint8_t *pixels, DCTELEM *block, int line_size);
     int (*sum_abs_dctelem)(DCTELEM *block/*align 16*/);
     /**
      * translational global motion compensation.
      */
     void (*gmc1)(uint8_t *dst/*align 8*/, uint8_t *src/*align 1*/, int srcStride, int h, int x16, int y16, int rounder);
     /**
      * global motion compensation.
@@ skipping 368 matching lines @@
 
     /* rv30 functions */
     qpel_mc_func put_rv30_tpel_pixels_tab[4][16];
     qpel_mc_func avg_rv30_tpel_pixels_tab[4][16];
 
     /* rv40 functions */
     qpel_mc_func put_rv40_qpel_pixels_tab[4][16];
     qpel_mc_func avg_rv40_qpel_pixels_tab[4][16];
     h264_chroma_mc_func put_rv40_chroma_pixels_tab[3];
     h264_chroma_mc_func avg_rv40_chroma_pixels_tab[3];
+
+    /* bink functions */
+    op_fill_func fill_block_tab[2];
+    void (*scale_block)(const uint8_t src[64]/*align 8*/, uint8_t *dst/*align 8*/, int linesize);
 } DSPContext;
 
 void dsputil_static_init(void);
 void dsputil_init(DSPContext* p, AVCodecContext *avctx);
 
 int ff_check_alignment(void);
 
 /**
  * permute block according to permuatation.
  * @param last last non zero element in scantable order
@@ skipping 53 matching lines @@
 void dsputil_init_alpha(DSPContext* c, AVCodecContext *avctx);
 void dsputil_init_arm(DSPContext* c, AVCodecContext *avctx);
 void dsputil_init_bfin(DSPContext* c, AVCodecContext *avctx);
 void dsputil_init_mlib(DSPContext* c, AVCodecContext *avctx);
 void dsputil_init_mmi(DSPContext* c, AVCodecContext *avctx);
 void dsputil_init_mmx(DSPContext* c, AVCodecContext *avctx);
 void dsputil_init_ppc(DSPContext* c, AVCodecContext *avctx);
 void dsputil_init_sh4(DSPContext* c, AVCodecContext *avctx);
 void dsputil_init_vis(DSPContext* c, AVCodecContext *avctx);
 
-#define DECLARE_ALIGNED_16(t, v, ...) DECLARE_ALIGNED(16, t, v)
-#define DECLARE_ALIGNED_8(t, v, ...)  DECLARE_ALIGNED(8, t, v)
+void ff_cavsdsp_init(DSPContext* c, AVCodecContext *avctx);
+void ff_rv30dsp_init(DSPContext* c, AVCodecContext *avctx);
+void ff_rv40dsp_init(DSPContext* c, AVCodecContext *avctx);
+void ff_vc1dsp_init(DSPContext* c, AVCodecContext *avctx);
+void ff_intrax8dsp_init(DSPContext* c, AVCodecContext *avctx);
+void ff_mlp_init(DSPContext* c, AVCodecContext *avctx);
+void ff_mlp_init_x86(DSPContext* c, AVCodecContext *avctx);
 
 #if HAVE_MMX
 
 #undef emms_c
 
 static inline void emms(void)
 {
     __asm__ volatile ("emms;":::"memory");
 }
 
@@ skipping 22 matching lines @@
 
 #define mm_flags 0
 #define mm_support() 0
 
 #endif
 
 #ifndef STRIDE_ALIGN
 #   define STRIDE_ALIGN 8
 #endif
 
+#define LOCAL_ALIGNED(a, t, v, s, ...)                          \
+    uint8_t la_##v[sizeof(t s __VA_ARGS__) + (a)];              \
+    t (*v) __VA_ARGS__ = (void *)FFALIGN((uintptr_t)la_##v, a)
+
+#if HAVE_LOCAL_ALIGNED_8
+#   define LOCAL_ALIGNED_8(t, v, s, ...) DECLARE_ALIGNED(8, t, v) s __VA_ARGS__
+#else
+#   define LOCAL_ALIGNED_8(t, v, s, ...) LOCAL_ALIGNED(8, t, v, s, __VA_ARGS__)
+#endif
+
+#if HAVE_LOCAL_ALIGNED_16
+#   define LOCAL_ALIGNED_16(t, v, s, ...) DECLARE_ALIGNED(16, t, v) s __VA_ARGS__
+#else
+#   define LOCAL_ALIGNED_16(t, v, s, ...) LOCAL_ALIGNED(16, t, v, s, __VA_ARGS__)
+#endif
+
 /* PSNR */
 void get_psnr(uint8_t *orig_image[3], uint8_t *coded_image[3],
               int orig_linesize[3], int coded_linesize,
               AVCodecContext *avctx);
 
-/* FFT computation */
-
-/* NOTE: soon integer code will be added, so you must use the
-   FFTSample type */
-typedef float FFTSample;
-
-typedef struct FFTComplex {
-    FFTSample re, im;
-} FFTComplex;
-
-typedef struct FFTContext {
-    int nbits;
-    int inverse;
-    uint16_t *revtab;
-    FFTComplex *exptab;
-    FFTComplex *exptab1; /* only used by SSE code */
-    FFTComplex *tmp_buf;
-    int mdct_size; /* size of MDCT (i.e. number of input data * 2) */
-    int mdct_bits; /* n = 2^nbits */
-    /* pre/post rotation tables */
-    FFTSample *tcos;
-    FFTSample *tsin;
-    void (*fft_permute)(struct FFTContext *s, FFTComplex *z);
-    void (*fft_calc)(struct FFTContext *s, FFTComplex *z);
-    void (*imdct_calc)(struct FFTContext *s, FFTSample *output, const FFTSample *input);
-    void (*imdct_half)(struct FFTContext *s, FFTSample *output, const FFTSample *input);
-    void (*mdct_calc)(struct FFTContext *s, FFTSample *output, const FFTSample *input);
-    int split_radix;
-    int permutation;
-#define FF_MDCT_PERM_NONE       0
-#define FF_MDCT_PERM_INTERLEAVE 1
-} FFTContext;
-
-#if CONFIG_HARDCODED_TABLES
-#define COSTABLE_CONST const
-#define SINTABLE_CONST const
-#define SINETABLE_CONST const
-#else
-#define COSTABLE_CONST
-#define SINTABLE_CONST
-#define SINETABLE_CONST
-#endif
-
-#define COSTABLE(size) \
-    COSTABLE_CONST DECLARE_ALIGNED_16(FFTSample, ff_cos_##size)[size/2]
-#define SINTABLE(size) \
-    SINTABLE_CONST DECLARE_ALIGNED_16(FFTSample, ff_sin_##size)[size/2]
-#define SINETABLE(size) \
-    SINETABLE_CONST DECLARE_ALIGNED_16(float, ff_sine_##size)[size]
-extern COSTABLE(16);
-extern COSTABLE(32);
-extern COSTABLE(64);
-extern COSTABLE(128);
-extern COSTABLE(256);
-extern COSTABLE(512);
-extern COSTABLE(1024);
-extern COSTABLE(2048);
-extern COSTABLE(4096);
-extern COSTABLE(8192);
-extern COSTABLE(16384);
-extern COSTABLE(32768);
-extern COSTABLE(65536);
-extern COSTABLE_CONST FFTSample* const ff_cos_tabs[17];
-
-/**
- * Initializes the cosine table in ff_cos_tabs[index]
- * \param index index in ff_cos_tabs array of the table to initialize
- */
-void ff_init_ff_cos_tabs(int index);
-
-extern SINTABLE(16);
-extern SINTABLE(32);
-extern SINTABLE(64);
-extern SINTABLE(128);
-extern SINTABLE(256);
-extern SINTABLE(512);
-extern SINTABLE(1024);
-extern SINTABLE(2048);
-extern SINTABLE(4096);
-extern SINTABLE(8192);
-extern SINTABLE(16384);
-extern SINTABLE(32768);
-extern SINTABLE(65536);
-
-/**
- * Sets up a complex FFT.
- * @param nbits           log2 of the length of the input array
- * @param inverse         if 0 perform the forward transform, if 1 perform the inverse
- */
-int ff_fft_init(FFTContext *s, int nbits, int inverse);
-void ff_fft_permute_c(FFTContext *s, FFTComplex *z);
-void ff_fft_calc_c(FFTContext *s, FFTComplex *z);
-
-void ff_fft_init_altivec(FFTContext *s);
-void ff_fft_init_mmx(FFTContext *s);
-void ff_fft_init_arm(FFTContext *s);
-
-/**
- * Do the permutation needed BEFORE calling ff_fft_calc().
- */
-static inline void ff_fft_permute(FFTContext *s, FFTComplex *z)
-{
-    s->fft_permute(s, z);
-}
-/**
- * Do a complex FFT with the parameters defined in ff_fft_init(). The
- * input data must be permuted before. No 1.0/sqrt(n) normalization is done.
- */
-static inline void ff_fft_calc(FFTContext *s, FFTComplex *z)
-{
-    s->fft_calc(s, z);
-}
-void ff_fft_end(FFTContext *s);
-
-/* MDCT computation */
-
-static inline void ff_imdct_calc(FFTContext *s, FFTSample *output, const FFTSample *input)
-{
-    s->imdct_calc(s, output, input);
-}
-static inline void ff_imdct_half(FFTContext *s, FFTSample *output, const FFTSample *input)
-{
-    s->imdct_half(s, output, input);
-}
-
-static inline void ff_mdct_calc(FFTContext *s, FFTSample *output,
-                                const FFTSample *input)
-{
-    s->mdct_calc(s, output, input);
-}
-
-/**
- * Generate a Kaiser-Bessel Derived Window.
- * @param   window  pointer to half window
- * @param   alpha   determines window shape
- * @param   n       size of half window
- */
-void ff_kbd_window_init(float *window, float alpha, int n);
-
-/**
- * Generate a sine window.
- * @param   window  pointer to half window
- * @param   n       size of half window
- */
-void ff_sine_window_init(float *window, int n);
-/**
- * initialize the specified entry of ff_sine_windows
- */
-void ff_init_ff_sine_windows(int index);
-extern SINETABLE(  32);
-extern SINETABLE(  64);
-extern SINETABLE( 128);
-extern SINETABLE( 256);
-extern SINETABLE( 512);
-extern SINETABLE(1024);
-extern SINETABLE(2048);
-extern SINETABLE(4096);
-extern SINETABLE_CONST float * const ff_sine_windows[13];
-
-int ff_mdct_init(FFTContext *s, int nbits, int inverse, double scale);
-void ff_imdct_calc_c(FFTContext *s, FFTSample *output, const FFTSample *input);
-void ff_imdct_half_c(FFTContext *s, FFTSample *output, const FFTSample *input);
-void ff_mdct_calc_c(FFTContext *s, FFTSample *output, const FFTSample *input);
-void ff_mdct_end(FFTContext *s);
-
-/* Real Discrete Fourier Transform */
-
-enum RDFTransformType {
-    RDFT,
-    IRDFT,
-    RIDFT,
-    IRIDFT,
-};
-
-typedef struct {
-    int nbits;
-    int inverse;
-    int sign_convention;
-
-    /* pre/post rotation tables */
-    const FFTSample *tcos;
-    SINTABLE_CONST FFTSample *tsin;
-    FFTContext fft;
-} RDFTContext;
-
-/**
- * Sets up a real FFT.
- * @param nbits           log2 of the length of the input array
- * @param trans           the type of transform
- */
-int ff_rdft_init(RDFTContext *s, int nbits, enum RDFTransformType trans);
-void ff_rdft_calc(RDFTContext *s, FFTSample *data);
-void ff_rdft_end(RDFTContext *s);
-
-/* Discrete Cosine Transform */
-
-typedef struct {
-    int nbits;
-    int inverse;
-    FFTSample *data;
-    RDFTContext rdft;
-    const float *costab;
-    FFTSample *csc2;
-} DCTContext;
-
-/**
- * Sets up (Inverse)DCT.
- * @param nbits           log2 of the length of the input array
- * @param inverse         >0 forward transform, <0 inverse transform
- */
-int  ff_dct_init(DCTContext *s, int nbits, int inverse);
-void ff_dct_calc(DCTContext *s, FFTSample *data);
-void ff_dct_end (DCTContext *s);
-
 #define WRAPPER8_16(name8, name16)\
 static int name16(void /*MpegEncContext*/ *s, uint8_t *dst, uint8_t *src, int stride, int h){\
     return name8(s, dst           , src           , stride, h)\
           +name8(s, dst+8         , src+8         , stride, h);\
 }
 
 #define WRAPPER8_16_SQ(name8, name16)\
 static int name16(void /*MpegEncContext*/ *s, uint8_t *dst, uint8_t *src, int stride, int h){\
     int score=0;\
     score +=name8(s, dst           , src           , stride, 8);\
@@ skipping 78 matching lines @@
         AV_WN32(dst+4 , AV_RN32(src+4 ));
         AV_WN32(dst+8 , AV_RN32(src+8 ));
         AV_WN32(dst+12, AV_RN32(src+12));
         dst[16]= src[16];
         dst+=dstStride;
         src+=srcStride;
     }
 }
 
 #endif /* AVCODEC_DSPUTIL_H */