X-Git-Url: https://git.sesse.net/?a=blobdiff_plain;f=libavcodec%2Fdsputil.c;h=c78fe40c7223c92615bb646debcee3936419d650;hb=dbc9f84ea6723c3d73dc9ec2aebacabc99a8d949;hp=91238578b6b2bc2d1209a88fab8ac9c0127fddb7;hpb=9d06037d48041ad8ccbae6c12aa9f3a313a89c4e;p=ffmpeg diff --git a/libavcodec/dsputil.c b/libavcodec/dsputil.c index 91238578b6b..c78fe40c722 100644 --- a/libavcodec/dsputil.c +++ b/libavcodec/dsputil.c @@ -38,7 +38,6 @@ #include "config.h" #include "ac3dec.h" #include "vorbis.h" -#include "png.h" uint8_t ff_cropTbl[256 + 2 * MAX_NEG_CROP] = {0, }; uint32_t ff_squareTbl[512] = {0, }; @@ -83,7 +82,7 @@ const uint8_t ff_zigzag248_direct[64] = { }; /* not permutated inverse zigzag_direct + 1 for MMX quantizer */ -DECLARE_ALIGNED(16, uint16_t, inv_zigzag_direct16)[64]; +DECLARE_ALIGNED(16, uint16_t, ff_inv_zigzag_direct16)[64]; const uint8_t ff_alternate_horizontal_scan[64] = { 0, 1, 2, 3, 8, 9, 16, 17, @@ -131,9 +130,6 @@ void ff_init_scantable(uint8_t *permutation, ScanTable *st, const uint8_t *src_s int j; j = src_scantable[i]; st->permutated[i] = permutation[j]; -#if ARCH_PPC - st->inverse[j] = i; -#endif } end=-1; @@ -363,65 +359,30 @@ static void diff_pixels_c(DCTELEM *restrict block, const uint8_t *s1, } -void ff_put_pixels_clamped_c(const DCTELEM *block, uint8_t *restrict pixels, - int line_size) -{ - int i; - uint8_t *cm = ff_cropTbl + MAX_NEG_CROP; - - /* read the pixels */ - for(i=0;i<8;i++) { - pixels[0] = cm[block[0]]; - pixels[1] = cm[block[1]]; - pixels[2] = cm[block[2]]; - pixels[3] = cm[block[3]]; - pixels[4] = cm[block[4]]; - pixels[5] = cm[block[5]]; - pixels[6] = cm[block[6]]; - pixels[7] = cm[block[7]]; - - pixels += line_size; - block += 8; - } -} - -static void put_pixels_clamped4_c(const DCTELEM *block, uint8_t *restrict pixels, - int line_size) -{ - int i; - uint8_t *cm = ff_cropTbl + MAX_NEG_CROP; - - /* read the pixels */ - for(i=0;i<4;i++) { - pixels[0] = cm[block[0]]; - pixels[1] = cm[block[1]]; - pixels[2] = cm[block[2]]; - pixels[3] = cm[block[3]]; - - pixels += line_size; - block += 8; - } -} - -static void put_pixels_clamped2_c(const DCTELEM *block, uint8_t *restrict pixels, +static void put_pixels_clamped_c(const DCTELEM *block, uint8_t *restrict pixels, int line_size) { int i; - uint8_t *cm = ff_cropTbl + MAX_NEG_CROP; /* read the pixels */ - for(i=0;i<2;i++) { - pixels[0] = cm[block[0]]; - pixels[1] = cm[block[1]]; + for(i=0;i<8;i++) { + pixels[0] = av_clip_uint8(block[0]); + pixels[1] = av_clip_uint8(block[1]); + pixels[2] = av_clip_uint8(block[2]); + pixels[3] = av_clip_uint8(block[3]); + pixels[4] = av_clip_uint8(block[4]); + pixels[5] = av_clip_uint8(block[5]); + pixels[6] = av_clip_uint8(block[6]); + pixels[7] = av_clip_uint8(block[7]); pixels += line_size; block += 8; } } -void ff_put_signed_pixels_clamped_c(const DCTELEM *block, - uint8_t *restrict pixels, - int line_size) +static void put_signed_pixels_clamped_c(const DCTELEM *block, + uint8_t *restrict pixels, + int line_size) { int i, j; @@ -440,54 +401,21 @@ void ff_put_signed_pixels_clamped_c(const DCTELEM *block, } } -void ff_add_pixels_clamped_c(const DCTELEM *block, uint8_t *restrict pixels, - int line_size) +static void add_pixels_clamped_c(const DCTELEM *block, uint8_t *restrict pixels, + int line_size) { int i; - uint8_t *cm = ff_cropTbl + MAX_NEG_CROP; /* read the pixels */ for(i=0;i<8;i++) { - pixels[0] = cm[pixels[0] + block[0]]; - pixels[1] = cm[pixels[1] + block[1]]; - pixels[2] = cm[pixels[2] + block[2]]; - pixels[3] = cm[pixels[3] + block[3]]; - pixels[4] = cm[pixels[4] + block[4]]; - pixels[5] = cm[pixels[5] + block[5]]; - pixels[6] = cm[pixels[6] + block[6]]; - pixels[7] = cm[pixels[7] + block[7]]; - pixels += line_size; - block += 8; - } -} - -static void add_pixels_clamped4_c(const DCTELEM *block, uint8_t *restrict pixels, - int line_size) -{ - int i; - uint8_t *cm = ff_cropTbl + MAX_NEG_CROP; - - /* read the pixels */ - for(i=0;i<4;i++) { - pixels[0] = cm[pixels[0] + block[0]]; - pixels[1] = cm[pixels[1] + block[1]]; - pixels[2] = cm[pixels[2] + block[2]]; - pixels[3] = cm[pixels[3] + block[3]]; - pixels += line_size; - block += 8; - } -} - -static void add_pixels_clamped2_c(const DCTELEM *block, uint8_t *restrict pixels, - int line_size) -{ - int i; - uint8_t *cm = ff_cropTbl + MAX_NEG_CROP; - - /* read the pixels */ - for(i=0;i<2;i++) { - pixels[0] = cm[pixels[0] + block[0]]; - pixels[1] = cm[pixels[1] + block[1]]; + pixels[0] = av_clip_uint8(pixels[0] + block[0]); + pixels[1] = av_clip_uint8(pixels[1] + block[1]); + pixels[2] = av_clip_uint8(pixels[2] + block[2]); + pixels[3] = av_clip_uint8(pixels[3] + block[3]); + pixels[4] = av_clip_uint8(pixels[4] + block[4]); + pixels[5] = av_clip_uint8(pixels[5] + block[5]); + pixels[6] = av_clip_uint8(pixels[6] + block[6]); + pixels[7] = av_clip_uint8(pixels[7] + block[7]); pixels += line_size; block += 8; } @@ -1779,7 +1707,7 @@ static void add_8x8basis_c(int16_t rem[64], int16_t basis[64], int scale){ } /** - * permutes an 8x8 block. + * Permute an 8x8 block. * @param block the block which will be permuted according to the given permutation vector * @param permutation the permutation vector * @param last the last non zero coefficient in scantable order, used to speed the permutation up @@ -1882,17 +1810,6 @@ static void add_bytes_c(uint8_t *dst, uint8_t *src, int w){ dst[i+0] += src[i+0]; } -static void add_bytes_l2_c(uint8_t *dst, uint8_t *src1, uint8_t *src2, int w){ - long i; - for(i=0; i<=w-sizeof(long); i+=sizeof(long)){ - long a = *(long*)(src1+i); - long b = *(long*)(src2+i); - *(long*)(dst+i) = ((a&pb_7f) + (b&pb_7f)) ^ ((a^b)&pb_80); - } - for(; i> shift; + res += *v1++ * *v2++; return res; } @@ -2703,63 +2606,28 @@ void ff_wmv2_idct_c(short * block){ static void ff_wmv2_idct_put_c(uint8_t *dest, int line_size, DCTELEM *block) { ff_wmv2_idct_c(block); - ff_put_pixels_clamped_c(block, dest, line_size); + put_pixels_clamped_c(block, dest, line_size); } static void ff_wmv2_idct_add_c(uint8_t *dest, int line_size, DCTELEM *block) { ff_wmv2_idct_c(block); - ff_add_pixels_clamped_c(block, dest, line_size); + add_pixels_clamped_c(block, dest, line_size); } static void ff_jref_idct_put(uint8_t *dest, int line_size, DCTELEM *block) { - j_rev_dct (block); - ff_put_pixels_clamped_c(block, dest, line_size); + ff_j_rev_dct (block); + put_pixels_clamped_c(block, dest, line_size); } static void ff_jref_idct_add(uint8_t *dest, int line_size, DCTELEM *block) { - j_rev_dct (block); - ff_add_pixels_clamped_c(block, dest, line_size); -} - -static void ff_jref_idct4_put(uint8_t *dest, int line_size, DCTELEM *block) -{ - j_rev_dct4 (block); - put_pixels_clamped4_c(block, dest, line_size); -} -static void ff_jref_idct4_add(uint8_t *dest, int line_size, DCTELEM *block) -{ - j_rev_dct4 (block); - add_pixels_clamped4_c(block, dest, line_size); -} - -static void ff_jref_idct2_put(uint8_t *dest, int line_size, DCTELEM *block) -{ - j_rev_dct2 (block); - put_pixels_clamped2_c(block, dest, line_size); -} -static void ff_jref_idct2_add(uint8_t *dest, int line_size, DCTELEM *block) -{ - j_rev_dct2 (block); - add_pixels_clamped2_c(block, dest, line_size); -} - -static void ff_jref_idct1_put(uint8_t *dest, int line_size, DCTELEM *block) -{ - uint8_t *cm = ff_cropTbl + MAX_NEG_CROP; - - dest[0] = cm[(block[0] + 4)>>3]; -} -static void ff_jref_idct1_add(uint8_t *dest, int line_size, DCTELEM *block) -{ - uint8_t *cm = ff_cropTbl + MAX_NEG_CROP; - - dest[0] = cm[dest[0] + ((block[0] + 4)>>3)]; + ff_j_rev_dct (block); + add_pixels_clamped_c(block, dest, line_size); } static void just_return(void *mem av_unused, int stride av_unused, int h av_unused) { return; } /* init static data */ -av_cold void dsputil_static_init(void) +av_cold void ff_dsputil_static_init(void) { int i; @@ -2773,7 +2641,7 @@ av_cold void dsputil_static_init(void) ff_squareTbl[i] = (i - 256) * (i - 256); } - for(i=0; i<64; i++) inv_zigzag_direct16[ff_zigzag_direct[i]]= i+1; + for(i=0; i<64; i++) ff_inv_zigzag_direct16[ff_zigzag_direct[i]]= i+1; } int ff_check_alignment(void){ @@ -2796,9 +2664,9 @@ int ff_check_alignment(void){ return 0; } -av_cold void dsputil_init(DSPContext* c, AVCodecContext *avctx) +av_cold void ff_dsputil_init(DSPContext* c, AVCodecContext *avctx) { - int i; + int i, j; ff_check_alignment(); @@ -2808,8 +2676,8 @@ av_cold void dsputil_init(DSPContext* c, AVCodecContext *avctx) c->fdct248 = ff_fdct248_islow_10; } else { if(avctx->dct_algo==FF_DCT_FASTINT) { - c->fdct = fdct_ifast; - c->fdct248 = fdct_ifast248; + c->fdct = ff_fdct_ifast; + c->fdct248 = ff_fdct_ifast248; } else if(avctx->dct_algo==FF_DCT_FAAN) { c->fdct = ff_faandct; @@ -2822,39 +2690,17 @@ av_cold void dsputil_init(DSPContext* c, AVCodecContext *avctx) } #endif //CONFIG_ENCODERS - if(avctx->lowres==1){ - c->idct_put= ff_jref_idct4_put; - c->idct_add= ff_jref_idct4_add; - c->idct = j_rev_dct4; - c->idct_permutation_type= FF_NO_IDCT_PERM; - }else if(avctx->lowres==2){ - c->idct_put= ff_jref_idct2_put; - c->idct_add= ff_jref_idct2_add; - c->idct = j_rev_dct2; - c->idct_permutation_type= FF_NO_IDCT_PERM; - }else if(avctx->lowres==3){ - c->idct_put= ff_jref_idct1_put; - c->idct_add= ff_jref_idct1_add; - c->idct = j_rev_dct1; - c->idct_permutation_type= FF_NO_IDCT_PERM; - }else{ - if (avctx->bits_per_raw_sample == 10) { - c->idct_put = ff_simple_idct_put_10; - c->idct_add = ff_simple_idct_add_10; - c->idct = ff_simple_idct_10; - c->idct_permutation_type = FF_NO_IDCT_PERM; - } else { + if (avctx->bits_per_raw_sample == 10) { + c->idct_put = ff_simple_idct_put_10; + c->idct_add = ff_simple_idct_add_10; + c->idct = ff_simple_idct_10; + c->idct_permutation_type = FF_NO_IDCT_PERM; + } else { if(avctx->idct_algo==FF_IDCT_INT){ c->idct_put= ff_jref_idct_put; c->idct_add= ff_jref_idct_add; - c->idct = j_rev_dct; + c->idct = ff_j_rev_dct; c->idct_permutation_type= FF_LIBMPEG2_IDCT_PERM; - }else if((CONFIG_VP3_DECODER || CONFIG_VP5_DECODER || CONFIG_VP6_DECODER ) && - avctx->idct_algo==FF_IDCT_VP3){ - c->idct_put= ff_vp3_idct_put_c; - c->idct_add= ff_vp3_idct_add_c; - c->idct = ff_vp3_idct_c; - c->idct_permutation_type= FF_NO_IDCT_PERM; }else if(avctx->idct_algo==FF_IDCT_WMV2){ c->idct_put= ff_wmv2_idct_put_c; c->idct_add= ff_wmv2_idct_add_c; @@ -2865,22 +2711,18 @@ av_cold void dsputil_init(DSPContext* c, AVCodecContext *avctx) c->idct_add= ff_faanidct_add; c->idct = ff_faanidct; c->idct_permutation_type= FF_NO_IDCT_PERM; - }else if(CONFIG_EATGQ_DECODER && avctx->idct_algo==FF_IDCT_EA) { - c->idct_put= ff_ea_idct_put_c; - c->idct_permutation_type= FF_NO_IDCT_PERM; }else{ //accurate/default c->idct_put = ff_simple_idct_put_8; c->idct_add = ff_simple_idct_add_8; c->idct = ff_simple_idct_8; c->idct_permutation_type= FF_NO_IDCT_PERM; } - } } c->diff_pixels = diff_pixels_c; - c->put_pixels_clamped = ff_put_pixels_clamped_c; - c->put_signed_pixels_clamped = ff_put_signed_pixels_clamped_c; - c->add_pixels_clamped = ff_add_pixels_clamped_c; + c->put_pixels_clamped = put_pixels_clamped_c; + c->put_signed_pixels_clamped = put_signed_pixels_clamped_c; + c->add_pixels_clamped = add_pixels_clamped_c; c->sum_abs_dctelem = sum_abs_dctelem_c; c->gmc1 = gmc1_c; c->gmc = ff_gmc_c; @@ -3003,7 +2845,6 @@ av_cold void dsputil_init(DSPContext* c, AVCodecContext *avctx) c->ssd_int8_vs_int16 = ssd_int8_vs_int16_c; c->add_bytes= add_bytes_c; - c->add_bytes_l2= add_bytes_l2_c; c->diff_bytes= diff_bytes_c; c->add_hfyu_median_prediction= add_hfyu_median_prediction_c; c->sub_hfyu_median_prediction= sub_hfyu_median_prediction_c; @@ -3011,33 +2852,23 @@ av_cold void dsputil_init(DSPContext* c, AVCodecContext *avctx) c->add_hfyu_left_prediction_bgr32 = add_hfyu_left_prediction_bgr32_c; c->bswap_buf= bswap_buf; c->bswap16_buf = bswap16_buf; -#if CONFIG_PNG_DECODER - c->add_png_paeth_prediction= ff_add_png_paeth_prediction; -#endif if (CONFIG_H263_DECODER || CONFIG_H263_ENCODER) { c->h263_h_loop_filter= h263_h_loop_filter_c; c->h263_v_loop_filter= h263_v_loop_filter_c; } - if (CONFIG_VP3_DECODER) { - c->vp3_h_loop_filter= ff_vp3_h_loop_filter_c; - c->vp3_v_loop_filter= ff_vp3_v_loop_filter_c; - c->vp3_idct_dc_add= ff_vp3_idct_dc_add_c; - } - c->h261_loop_filter= h261_loop_filter_c; c->try_8x8basis= try_8x8basis_c; c->add_8x8basis= add_8x8basis_c; #if CONFIG_VORBIS_DECODER - c->vorbis_inverse_coupling = vorbis_inverse_coupling; + c->vorbis_inverse_coupling = ff_vorbis_inverse_coupling; #endif #if CONFIG_AC3_DECODER c->ac3_downmix = ff_ac3_downmix_c; #endif - c->vector_fmul = vector_fmul_c; c->vector_fmul_reverse = vector_fmul_reverse_c; c->vector_fmul_add = vector_fmul_add_c; c->vector_fmul_window = vector_fmul_window_c; @@ -3046,11 +2877,10 @@ av_cold void dsputil_init(DSPContext* c, AVCodecContext *avctx) c->scalarproduct_and_madd_int16 = scalarproduct_and_madd_int16_c; c->apply_window_int16 = apply_window_int16_c; c->vector_clip_int32 = vector_clip_int32_c; - c->scalarproduct_float = scalarproduct_float_c; + c->scalarproduct_float = ff_scalarproduct_float_c; c->butterflies_float = butterflies_float_c; c->butterflies_float_interleave = butterflies_float_interleave_c; c->vector_fmul_scalar = vector_fmul_scalar_c; - c->vector_fmac_scalar = vector_fmac_scalar_c; c->shrink[0]= av_image_copy_plane; c->shrink[1]= ff_shrink22; @@ -3147,28 +2977,29 @@ av_cold void dsputil_init(DSPContext* c, AVCodecContext *avctx) } break; default: - av_log(avctx, AV_LOG_DEBUG, "Unsupported bit depth: %d\n", avctx->bits_per_raw_sample); - case 8: BIT_DEPTH_FUNCS(8, _16); break; } - if (HAVE_MMX) dsputil_init_mmx (c, avctx); - if (ARCH_ARM) dsputil_init_arm (c, avctx); - if (CONFIG_MLIB) dsputil_init_mlib (c, avctx); - if (HAVE_VIS) dsputil_init_vis (c, avctx); - if (ARCH_ALPHA) dsputil_init_alpha (c, avctx); - if (ARCH_PPC) dsputil_init_ppc (c, avctx); - if (HAVE_MMI) dsputil_init_mmi (c, avctx); - if (ARCH_SH4) dsputil_init_sh4 (c, avctx); - if (ARCH_BFIN) dsputil_init_bfin (c, avctx); - - for(i=0; i<64; i++){ - if(!c->put_2tap_qpel_pixels_tab[0][i]) - c->put_2tap_qpel_pixels_tab[0][i]= c->put_h264_qpel_pixels_tab[0][i]; - if(!c->avg_2tap_qpel_pixels_tab[0][i]) - c->avg_2tap_qpel_pixels_tab[0][i]= c->avg_h264_qpel_pixels_tab[0][i]; + if (HAVE_MMX) ff_dsputil_init_mmx (c, avctx); + if (ARCH_ARM) ff_dsputil_init_arm (c, avctx); + if (HAVE_VIS) ff_dsputil_init_vis (c, avctx); + if (ARCH_ALPHA) ff_dsputil_init_alpha (c, avctx); + if (ARCH_PPC) ff_dsputil_init_ppc (c, avctx); + if (HAVE_MMI) ff_dsputil_init_mmi (c, avctx); + if (ARCH_SH4) ff_dsputil_init_sh4 (c, avctx); + if (ARCH_BFIN) ff_dsputil_init_bfin (c, avctx); + + for (i = 0; i < 4; i++) { + for (j = 0; j < 16; j++) { + if(!c->put_2tap_qpel_pixels_tab[i][j]) + c->put_2tap_qpel_pixels_tab[i][j] = + c->put_h264_qpel_pixels_tab[i][j]; + if(!c->avg_2tap_qpel_pixels_tab[i][j]) + c->avg_2tap_qpel_pixels_tab[i][j] = + c->avg_h264_qpel_pixels_tab[i][j]; + } } ff_init_scantable_permutation(c->idct_permutation,