X-Git-Url: https://git.sesse.net/?a=blobdiff_plain;f=common%2Fframe.c;h=cc9c96ba8d94af0b7680d8109034241586e296a4;hb=cbb90707e443f0da2521bda1b98cab5705451b8f;hp=96291f043f66f8baf84438a91c10fa7c2bc2c680;hpb=2f2ab0fa6c873c32363d7c3115f483fafdbe326f;p=x264 diff --git a/common/frame.c b/common/frame.c index 96291f04..cc9c96ba 100644 --- a/common/frame.c +++ b/common/frame.c @@ -1,7 +1,7 @@ /***************************************************************************** * frame.c: frame handling ***************************************************************************** - * Copyright (C) 2003-2010 x264 project + * Copyright (C) 2003-2012 x264 project * * Authors: Laurent Aimar * Loren Merritt @@ -42,14 +42,36 @@ static int align_plane_size( int x, int disalign ) return x; } -x264_frame_t *x264_frame_new( x264_t *h, int b_fdec ) +static int x264_frame_internal_csp( int external_csp ) { - x264_frame_t *frame; + switch( external_csp & X264_CSP_MASK ) + { + case X264_CSP_NV12: + case X264_CSP_I420: + case X264_CSP_YV12: + return X264_CSP_NV12; + case X264_CSP_NV16: + case X264_CSP_I422: + case X264_CSP_YV16: + return X264_CSP_NV16; + case X264_CSP_I444: + case X264_CSP_YV24: + case X264_CSP_BGR: + case X264_CSP_BGRA: + case X264_CSP_RGB: + return X264_CSP_I444; + default: + return X264_CSP_NONE; + } +} +static x264_frame_t *x264_frame_new( x264_t *h, int b_fdec ) +{ + x264_frame_t *frame; + int i_csp = x264_frame_internal_csp( h->param.i_csp ); int i_mb_count = h->mb.i_mb_count; - int i_stride, i_width, i_lines; - int i_padv = PADV << h->param.b_interlaced; - int luma_plane_size, chroma_plane_size; + int i_stride, i_width, i_lines, luma_plane_count; + int i_padv = PADV << PARAM_INTERLACED; int align = h->param.cpu&X264_CPU_CACHELINE_64 ? 64 : h->param.cpu&X264_CPU_CACHELINE_32 ? 32 : 16; int disalign = h->param.cpu&X264_CPU_ALTIVEC ? 1<<9 : 1<<10; @@ -60,14 +82,32 @@ x264_frame_t *x264_frame_new( x264_t *h, int b_fdec ) i_lines = h->mb.i_mb_height*16; i_stride = align_stride( i_width + 2*PADH, align, disalign ); - frame->i_plane = 2; - for( int i = 0; i < 2; i++ ) + if( i_csp == X264_CSP_NV12 || i_csp == X264_CSP_NV16 ) { - frame->i_width[i] = i_width >> i; - frame->i_lines[i] = i_lines >> i; - frame->i_stride[i] = i_stride; + luma_plane_count = 1; + frame->i_plane = 2; + for( int i = 0; i < 2; i++ ) + { + frame->i_width[i] = i_width >> i; + frame->i_lines[i] = i_lines >> (i && i_csp == X264_CSP_NV12); + frame->i_stride[i] = i_stride; + } } + else if( i_csp == X264_CSP_I444 ) + { + luma_plane_count = 3; + frame->i_plane = 3; + for( int i = 0; i < 3; i++ ) + { + frame->i_width[i] = i_width; + frame->i_lines[i] = i_lines; + frame->i_stride[i] = i_stride; + } + } + else + goto fail; + frame->i_csp = i_csp; frame->i_width_lowres = frame->i_width[0]/2; frame->i_lines_lowres = frame->i_lines[0]/2; frame->i_stride_lowres = align_stride( frame->i_width_lowres + 2*PADH, align, disalign<<1 ); @@ -95,25 +135,47 @@ x264_frame_t *x264_frame_new( x264_t *h, int b_fdec ) frame->orig = frame; - luma_plane_size = align_plane_size( frame->i_stride[0] * (frame->i_lines[0] + 2*i_padv), disalign ); - chroma_plane_size = (frame->i_stride[1] * (frame->i_lines[1] + i_padv)); - - CHECKED_MALLOC( frame->buffer[1], chroma_plane_size * sizeof(pixel) ); - frame->plane[1] = frame->buffer[1] + frame->i_stride[1] * i_padv/2 + PADH; + if( i_csp == X264_CSP_NV12 || i_csp == X264_CSP_NV16 ) + { + int chroma_padv = i_padv >> (i_csp == X264_CSP_NV12); + int chroma_plane_size = (frame->i_stride[1] * (frame->i_lines[1] + 2*chroma_padv)); + CHECKED_MALLOC( frame->buffer[1], chroma_plane_size * sizeof(pixel) ); + frame->plane[1] = frame->buffer[1] + frame->i_stride[1] * chroma_padv + PADH; + if( PARAM_INTERLACED ) + { + CHECKED_MALLOC( frame->buffer_fld[1], chroma_plane_size * sizeof(pixel) ); + frame->plane_fld[1] = frame->buffer_fld[1] + frame->i_stride[1] * chroma_padv + PADH; + } + } /* all 4 luma planes allocated together, since the cacheline split code * requires them to be in-phase wrt cacheline alignment. */ - if( h->param.analyse.i_subpel_refine && b_fdec ) - { - CHECKED_MALLOC( frame->buffer[0], 4*luma_plane_size * sizeof(pixel) ); - for( int i = 0; i < 4; i++ ) - frame->filtered[i] = frame->buffer[0] + i*luma_plane_size + frame->i_stride[0] * i_padv + PADH; - frame->plane[0] = frame->filtered[0]; - } - else + + for( int p = 0; p < luma_plane_count; p++ ) { - CHECKED_MALLOC( frame->buffer[0], luma_plane_size * sizeof(pixel) ); - frame->filtered[0] = frame->plane[0] = frame->buffer[0] + frame->i_stride[0] * i_padv + PADH; + int luma_plane_size = align_plane_size( frame->i_stride[p] * (frame->i_lines[p] + 2*i_padv), disalign ); + if( h->param.analyse.i_subpel_refine && b_fdec ) + { + /* FIXME: Don't allocate both buffers in non-adaptive MBAFF. */ + CHECKED_MALLOC( frame->buffer[p], 4*luma_plane_size * sizeof(pixel) ); + if( PARAM_INTERLACED ) + CHECKED_MALLOC( frame->buffer_fld[p], 4*luma_plane_size * sizeof(pixel) ); + for( int i = 0; i < 4; i++ ) + { + frame->filtered[p][i] = frame->buffer[p] + i*luma_plane_size + frame->i_stride[p] * i_padv + PADH; + frame->filtered_fld[p][i] = frame->buffer_fld[p] + i*luma_plane_size + frame->i_stride[p] * i_padv + PADH; + } + frame->plane[p] = frame->filtered[p][0]; + frame->plane_fld[p] = frame->filtered_fld[p][0]; + } + else + { + CHECKED_MALLOC( frame->buffer[p], luma_plane_size * sizeof(pixel) ); + if( PARAM_INTERLACED ) + CHECKED_MALLOC( frame->buffer_fld[p], luma_plane_size * sizeof(pixel) ); + frame->filtered[p][0] = frame->plane[p] = frame->buffer[p] + frame->i_stride[p] * i_padv + PADH; + frame->filtered_fld[p][0] = frame->plane_fld[p] = frame->buffer_fld[p] + frame->i_stride[p] * i_padv + PADH; + } } frame->b_duplicate = 0; @@ -139,18 +201,23 @@ x264_frame_t *x264_frame_new( x264_t *h, int b_fdec ) } CHECKED_MALLOC( frame->i_row_bits, i_lines/16 * sizeof(int) ); CHECKED_MALLOC( frame->f_row_qp, i_lines/16 * sizeof(float) ); + CHECKED_MALLOC( frame->f_row_qscale, i_lines/16 * sizeof(float) ); if( h->param.analyse.i_me_method >= X264_ME_ESA ) { CHECKED_MALLOC( frame->buffer[3], frame->i_stride[0] * (frame->i_lines[0] + 2*i_padv) * sizeof(uint16_t) << h->frames.b_have_sub8x8_esa ); frame->integral = (uint16_t*)frame->buffer[3] + frame->i_stride[0] * i_padv + PADH; } + if( PARAM_INTERLACED ) + CHECKED_MALLOC( frame->field, i_mb_count * sizeof(uint8_t) ); + if( h->param.analyse.b_mb_info ) + CHECKED_MALLOC( frame->effective_qp, i_mb_count * sizeof(uint8_t) ); } else /* fenc frame */ { if( h->frames.b_have_lowres ) { - luma_plane_size = align_plane_size( frame->i_stride_lowres * (frame->i_lines[0]/2 + 2*PADV), disalign ); + int luma_plane_size = align_plane_size( frame->i_stride_lowres * (frame->i_lines[0]/2 + 2*PADV), disalign ); CHECKED_MALLOC( frame->buffer_lowres[0], 4 * luma_plane_size * sizeof(pixel) ); for( int i = 0; i < 4; i++ ) @@ -162,7 +229,7 @@ x264_frame_t *x264_frame_new( x264_t *h, int b_fdec ) CHECKED_MALLOCZERO( frame->lowres_mvs[j][i], 2*h->mb.i_mb_count*sizeof(int16_t) ); CHECKED_MALLOC( frame->lowres_mv_costs[j][i], h->mb.i_mb_count*sizeof(int) ); } - CHECKED_MALLOC( frame->i_propagate_cost, (i_mb_count+3) * sizeof(uint16_t) ); + CHECKED_MALLOC( frame->i_propagate_cost, (i_mb_count+7) * sizeof(uint16_t) ); for( int j = 0; j <= h->param.i_bframe+1; j++ ) for( int i = 0; i <= h->param.i_bframe+1; i++ ) CHECKED_MALLOC( frame->lowres_costs[j][i], (i_mb_count+3) * sizeof(uint16_t) ); @@ -198,7 +265,10 @@ void x264_frame_delete( x264_frame_t *frame ) if( !frame->b_duplicate ) { for( int i = 0; i < 4; i++ ) + { x264_free( frame->buffer[i] ); + x264_free( frame->buffer_fld[i] ); + } for( int i = 0; i < 4; i++ ) x264_free( frame->buffer_lowres[i] ); for( int i = 0; i < X264_BFRAME_MAX+2; i++ ) @@ -219,6 +289,9 @@ void x264_frame_delete( x264_frame_t *frame ) x264_free( frame->i_inv_qscale_factor ); x264_free( frame->i_row_bits ); x264_free( frame->f_row_qp ); + x264_free( frame->f_row_qscale ); + x264_free( frame->field ); + x264_free( frame->effective_qp ); x264_free( frame->mb_type ); x264_free( frame->mb_partition ); x264_free( frame->mv[0] ); @@ -227,6 +300,16 @@ void x264_frame_delete( x264_frame_t *frame ) x264_free( frame->mv16x16-1 ); x264_free( frame->ref[0] ); x264_free( frame->ref[1] ); + if( frame->param && frame->param->param_free ) + frame->param->param_free( frame->param ); + if( frame->mb_info_free ) + frame->mb_info_free( frame->mb_info ); + if( frame->extra_sei.sei_free ) + { + for( int i = 0; i < frame->extra_sei.num_payloads; i++ ) + frame->extra_sei.sei_free( frame->extra_sei.payloads[i].payload ); + frame->extra_sei.sei_free( frame->extra_sei.payloads ); + } x264_pthread_mutex_destroy( &frame->mutex ); x264_pthread_cond_destroy( &frame->cv ); } @@ -257,13 +340,13 @@ static int get_plane_ptr( x264_t *h, x264_picture_t *src, uint8_t **pix, int *st int x264_frame_copy_picture( x264_t *h, x264_frame_t *dst, x264_picture_t *src ) { int i_csp = src->img.i_csp & X264_CSP_MASK; - if( i_csp <= X264_CSP_NONE || i_csp >= X264_CSP_MAX ) + if( dst->i_csp != x264_frame_internal_csp( i_csp ) ) { x264_log( h, X264_LOG_ERROR, "Invalid input colorspace\n" ); return -1; } -#if X264_HIGH_BIT_DEPTH +#if HIGH_BIT_DEPTH if( !(src->img.i_csp & X264_CSP_HIGH_DEPTH) ) { x264_log( h, X264_LOG_ERROR, "This build of x264 requires high depth input. Rebuild to support 8-bit input.\n" ); @@ -283,25 +366,58 @@ int x264_frame_copy_picture( x264_t *h, x264_frame_t *dst, x264_picture_t *src ) dst->param = src->param; dst->i_pic_struct = src->i_pic_struct; dst->extra_sei = src->extra_sei; + dst->opaque = src->opaque; + dst->mb_info = h->param.analyse.b_mb_info ? src->prop.mb_info : NULL; + dst->mb_info_free = h->param.analyse.b_mb_info ? src->prop.mb_info_free : NULL; uint8_t *pix[3]; int stride[3]; - get_plane_ptr( h, src, &pix[0], &stride[0], 0, 0, 0 ); - h->mc.plane_copy( dst->plane[0], dst->i_stride[0], pix[0], stride[0], - h->param.i_width, h->param.i_height ); - if( i_csp == X264_CSP_NV12 ) + if ( i_csp >= X264_CSP_BGR ) { - get_plane_ptr( h, src, &pix[1], &stride[1], 1, 0, 1 ); - h->mc.plane_copy( dst->plane[1], dst->i_stride[1], pix[1], stride[1], - h->param.i_width, h->param.i_height>>1 ); + stride[0] = src->img.i_stride[0]; + pix[0] = src->img.plane[0]; + if( src->img.i_csp & X264_CSP_VFLIP ) + { + pix[0] += (h->param.i_height-1) * stride[0]; + stride[0] = -stride[0]; + } + int b = i_csp==X264_CSP_RGB; + h->mc.plane_copy_deinterleave_rgb( dst->plane[1+b], dst->i_stride[1+b], + dst->plane[0], dst->i_stride[0], + dst->plane[2-b], dst->i_stride[2-b], + (pixel*)pix[0], stride[0]/sizeof(pixel), i_csp==X264_CSP_BGRA ? 4 : 3, h->param.i_width, h->param.i_height ); } else { - get_plane_ptr( h, src, &pix[1], &stride[1], i_csp==X264_CSP_I420 ? 1 : 2, 1, 1 ); - get_plane_ptr( h, src, &pix[2], &stride[2], i_csp==X264_CSP_I420 ? 2 : 1, 1, 1 ); - h->mc.plane_copy_interleave( dst->plane[1], dst->i_stride[1], - pix[1], stride[1], pix[2], stride[2], - h->param.i_width>>1, h->param.i_height>>1 ); + int v_shift = CHROMA_V_SHIFT; + get_plane_ptr( h, src, &pix[0], &stride[0], 0, 0, 0 ); + h->mc.plane_copy( dst->plane[0], dst->i_stride[0], (pixel*)pix[0], + stride[0]/sizeof(pixel), h->param.i_width, h->param.i_height ); + if( i_csp == X264_CSP_NV12 || i_csp == X264_CSP_NV16 ) + { + get_plane_ptr( h, src, &pix[1], &stride[1], 1, 0, v_shift ); + h->mc.plane_copy( dst->plane[1], dst->i_stride[1], (pixel*)pix[1], + stride[1]/sizeof(pixel), h->param.i_width, h->param.i_height>>v_shift ); + } + else if( i_csp == X264_CSP_I420 || i_csp == X264_CSP_I422 || i_csp == X264_CSP_YV12 || i_csp == X264_CSP_YV16 ) + { + int uv_swap = i_csp == X264_CSP_YV12 || i_csp == X264_CSP_YV16; + get_plane_ptr( h, src, &pix[1], &stride[1], uv_swap ? 2 : 1, 1, v_shift ); + get_plane_ptr( h, src, &pix[2], &stride[2], uv_swap ? 1 : 2, 1, v_shift ); + h->mc.plane_copy_interleave( dst->plane[1], dst->i_stride[1], + (pixel*)pix[1], stride[1]/sizeof(pixel), + (pixel*)pix[2], stride[2]/sizeof(pixel), + h->param.i_width>>1, h->param.i_height>>v_shift ); + } + else //if( i_csp == X264_CSP_I444 || i_csp == X264_CSP_YV24 ) + { + get_plane_ptr( h, src, &pix[1], &stride[1], i_csp==X264_CSP_I444 ? 1 : 2, 0, 0 ); + get_plane_ptr( h, src, &pix[2], &stride[2], i_csp==X264_CSP_I444 ? 2 : 1, 0, 0 ); + h->mc.plane_copy( dst->plane[1], dst->i_stride[1], (pixel*)pix[1], + stride[1]/sizeof(pixel), h->param.i_width, h->param.i_height ); + h->mc.plane_copy( dst->plane[2], dst->i_stride[2], (pixel*)pix[2], + stride[2]/sizeof(pixel), h->param.i_width, h->param.i_height ); + } } return 0; } @@ -309,23 +425,56 @@ int x264_frame_copy_picture( x264_t *h, x264_frame_t *dst, x264_picture_t *src ) static void ALWAYS_INLINE pixel_memset( pixel *dst, pixel *src, int len, int size ) { uint8_t *dstp = (uint8_t*)dst; - if( size == 1 ) - memset(dst, *src, len); - else if( size == 2 ) + uint32_t v1 = *src; + uint32_t v2 = size == 1 ? v1 + (v1 << 8) : M16( src ); + uint32_t v4 = size <= 2 ? v2 + (v2 << 16) : M32( src ); + int i = 0; + len *= size; + + /* Align the input pointer if it isn't already */ + if( (intptr_t)dstp & (WORD_SIZE - 1) ) { - int v = M16( src ); - for( int i = 0; i < len; i++ ) - M16( dstp+i*2 ) = v; + if( size <= 2 && ((intptr_t)dstp & 3) ) + { + if( size == 1 && ((intptr_t)dstp & 1) ) + dstp[i++] = v1; + if( (intptr_t)dstp & 2 ) + { + M16( dstp+i ) = v2; + i += 2; + } + } + if( WORD_SIZE == 8 && (intptr_t)dstp & 4 ) + { + M32( dstp+i ) = v4; + i += 4; + } } - else if( size == 4 ) + + /* Main copy loop */ + if( WORD_SIZE == 8 ) { - int v = M32( src ); - for( int i = 0; i < len; i++ ) - M32( dstp+i*4 ) = v; + uint64_t v8 = v4 + ((uint64_t)v4<<32); + for( ; i < len - 7; i+=8 ) + M64( dstp+i ) = v8; + } + for( ; i < len - 3; i+=4 ) + M32( dstp+i ) = v4; + + /* Finish up the last few bytes */ + if( size <= 2 ) + { + if( i < len - 1 ) + { + M16( dstp+i ) = v2; + i += 2; + } + if( size == 1 && i != len ) + dstp[i] = v1; } } -static void plane_expand_border( pixel *pix, int i_stride, int i_width, int i_height, int i_padh, int i_padv, int b_pad_top, int b_pad_bottom, int b_chroma ) +static void ALWAYS_INLINE plane_expand_border( pixel *pix, int i_stride, int i_width, int i_height, int i_padh, int i_padv, int b_pad_top, int b_pad_bottom, int b_chroma ) { #define PPIXEL(x, y) ( pix + (x) + (y)*i_stride ) for( int y = 0; y < i_height; y++ ) @@ -346,30 +495,45 @@ static void plane_expand_border( pixel *pix, int i_stride, int i_width, int i_he #undef PPIXEL } -void x264_frame_expand_border( x264_t *h, x264_frame_t *frame, int mb_y, int b_end ) +void x264_frame_expand_border( x264_t *h, x264_frame_t *frame, int mb_y ) { - int b_start = !mb_y; - if( mb_y & h->sh.b_mbaff ) + int pad_top = mb_y == 0; + int pad_bot = mb_y == h->mb.i_mb_height - (1 << SLICE_MBAFF); + int b_start = mb_y == h->i_threadslice_start; + int b_end = mb_y == h->i_threadslice_end - (1 << SLICE_MBAFF); + if( mb_y & SLICE_MBAFF ) return; for( int i = 0; i < frame->i_plane; i++ ) { + int h_shift = i && CHROMA_H_SHIFT; + int v_shift = i && CHROMA_V_SHIFT; int stride = frame->i_stride[i]; - int width = 16*h->sps->i_mb_width; - int height = (b_end ? 16*(h->mb.i_mb_height - mb_y) >> h->sh.b_mbaff : 16) >> !!i; + int width = 16*h->mb.i_mb_width; + int height = (pad_bot ? 16*(h->mb.i_mb_height - mb_y) >> SLICE_MBAFF : 16) >> v_shift; int padh = PADH; - int padv = PADV >> !!i; + int padv = PADV >> v_shift; // buffer: 2 chroma, 3 luma (rounded to 4) because deblocking goes beyond the top of the mb - pixel *pix = frame->plane[i] + X264_MAX(0, (16*mb_y-4)*stride >> !!i); if( b_end && !b_start ) - height += 4 >> (!!i + h->sh.b_mbaff); - if( h->sh.b_mbaff ) + height += 4 >> (v_shift + SLICE_MBAFF); + pixel *pix; + int starty = 16*mb_y - 4*!b_start; + if( SLICE_MBAFF ) { - plane_expand_border( pix, stride*2, width, height, padh, padv, b_start, b_end, i ); - plane_expand_border( pix+stride, stride*2, width, height, padh, padv, b_start, b_end, i ); + // border samples for each field are extended separately + pix = frame->plane_fld[i] + (starty*stride >> v_shift); + plane_expand_border( pix, stride*2, width, height, padh, padv, pad_top, pad_bot, h_shift ); + plane_expand_border( pix+stride, stride*2, width, height, padh, padv, pad_top, pad_bot, h_shift ); + + height = (pad_bot ? 16*(h->mb.i_mb_height - mb_y) : 32) >> v_shift; + if( b_end && !b_start ) + height += 4 >> v_shift; + pix = frame->plane[i] + (starty*stride >> v_shift); + plane_expand_border( pix, stride, width, height, padh, padv, pad_top, pad_bot, h_shift ); } else { - plane_expand_border( pix, stride, width, height, padh, padv, b_start, b_end, i ); + pix = frame->plane[i] + (starty*stride >> v_shift); + plane_expand_border( pix, stride, width, height, padh, padv, pad_top, pad_bot, h_shift ); } } } @@ -380,23 +544,26 @@ void x264_frame_expand_border_filtered( x264_t *h, x264_frame_t *frame, int mb_y * but up to 3 of the horizontal ones may be wrong. we want to expand border from the last filtered pixel */ int b_start = !mb_y; - int stride = frame->i_stride[0]; int width = 16*h->mb.i_mb_width + 8; - int height = b_end ? (16*(h->mb.i_mb_height - mb_y) >> h->sh.b_mbaff) + 16 : 16; + int height = b_end ? (16*(h->mb.i_mb_height - mb_y) >> SLICE_MBAFF) + 16 : 16; int padh = PADH - 4; int padv = PADV - 8; - for( int i = 1; i < 4; i++ ) - { - // buffer: 8 luma, to match the hpel filter - pixel *pix = frame->filtered[i] + (16*mb_y - (8 << h->sh.b_mbaff)) * stride - 4; - if( h->sh.b_mbaff ) + for( int p = 0; p < (CHROMA444 ? 3 : 1); p++ ) + for( int i = 1; i < 4; i++ ) { - plane_expand_border( pix, stride*2, width, height, padh, padv, b_start, b_end, 0 ); - plane_expand_border( pix+stride, stride*2, width, height, padh, padv, b_start, b_end, 0 ); + int stride = frame->i_stride[p]; + // buffer: 8 luma, to match the hpel filter + pixel *pix; + if( SLICE_MBAFF ) + { + pix = frame->filtered_fld[p][i] + (16*mb_y - 16) * stride - 4; + plane_expand_border( pix, stride*2, width, height, padh, padv, b_start, b_end, 0 ); + plane_expand_border( pix+stride, stride*2, width, height, padh, padv, b_start, b_end, 0 ); + } + + pix = frame->filtered[p][i] + (16*mb_y - 8) * stride - 4; + plane_expand_border( pix, stride, width, height << SLICE_MBAFF, padh, padv, b_start, b_end, 0 ); } - else - plane_expand_border( pix, stride, width, height, padh, padv, b_start, b_end, 0 ); - } } void x264_frame_expand_border_lowres( x264_frame_t *frame ) @@ -405,32 +572,55 @@ void x264_frame_expand_border_lowres( x264_frame_t *frame ) plane_expand_border( frame->lowres[i], frame->i_stride_lowres, frame->i_width_lowres, frame->i_lines_lowres, PADH, PADV, 1, 1, 0 ); } +void x264_frame_expand_border_chroma( x264_t *h, x264_frame_t *frame, int plane ) +{ + int v_shift = CHROMA_V_SHIFT; + plane_expand_border( frame->plane[plane], frame->i_stride[plane], 16*h->mb.i_mb_width, 16*h->mb.i_mb_height>>v_shift, + PADH, PADV>>v_shift, 1, 1, CHROMA_H_SHIFT ); +} + void x264_frame_expand_border_mod16( x264_t *h, x264_frame_t *frame ) { for( int i = 0; i < frame->i_plane; i++ ) { int i_width = h->param.i_width; - int i_height = h->param.i_height >> !!i; + int h_shift = i && CHROMA_H_SHIFT; + int v_shift = i && CHROMA_V_SHIFT; + int i_height = h->param.i_height >> v_shift; int i_padx = (h->mb.i_mb_width * 16 - h->param.i_width); - int i_pady = (h->mb.i_mb_height * 16 - h->param.i_height) >> !!i; + int i_pady = (h->mb.i_mb_height * 16 - h->param.i_height) >> v_shift; if( i_padx ) { for( int y = 0; y < i_height; y++ ) pixel_memset( &frame->plane[i][y*frame->i_stride[i] + i_width], - &frame->plane[i][y*frame->i_stride[i] + i_width - 1-i], - i_padx>>i, sizeof(pixel)<plane[i][y*frame->i_stride[i] + i_width - 1-h_shift], + i_padx>>h_shift, sizeof(pixel)<plane[i][y*frame->i_stride[i]], - &frame->plane[i][(i_height-(~y&h->param.b_interlaced)-1)*frame->i_stride[i]], + &frame->plane[i][(i_height-(~y&PARAM_INTERLACED)-1)*frame->i_stride[i]], (i_width + i_padx) * sizeof(pixel) ); } } } +void x264_expand_border_mbpair( x264_t *h, int mb_x, int mb_y ) +{ + for( int i = 0; i < h->fenc->i_plane; i++ ) + { + int v_shift = i && CHROMA_V_SHIFT; + int stride = h->fenc->i_stride[i]; + int height = h->param.i_height >> v_shift; + int pady = (h->mb.i_mb_height * 16 - h->param.i_height) >> v_shift; + pixel *fenc = h->fenc->plane[i] + 16*mb_x; + for( int y = height; y < height + pady; y++ ) + memcpy( fenc + y*stride, fenc + (height-1)*stride, 16*sizeof(pixel) ); + } +} + /* threading */ void x264_frame_cond_broadcast( x264_frame_t *frame, int i_lines_completed ) { @@ -448,6 +638,23 @@ void x264_frame_cond_wait( x264_frame_t *frame, int i_lines_completed ) x264_pthread_mutex_unlock( &frame->mutex ); } +void x264_threadslice_cond_broadcast( x264_t *h, int pass ) +{ + x264_pthread_mutex_lock( &h->mutex ); + h->i_threadslice_pass = pass; + if( pass > 0 ) + x264_pthread_cond_broadcast( &h->cv ); + x264_pthread_mutex_unlock( &h->mutex ); +} + +void x264_threadslice_cond_wait( x264_t *h, int pass ) +{ + x264_pthread_mutex_lock( &h->mutex ); + while( h->i_threadslice_pass < pass ) + x264_pthread_cond_wait( &h->cv, &h->mutex ); + x264_pthread_mutex_unlock( &h->mutex ); +} + /* list operators */ void x264_frame_push( x264_frame_t **list, x264_frame_t *frame ) @@ -539,35 +746,18 @@ x264_frame_t *x264_frame_pop_blank_unused( x264_t *h ) return frame; } -void x264_frame_sort( x264_frame_t **list, int b_dts ) -{ - int b_ok; - do { - b_ok = 1; - for( int i = 0; list[i+1]; i++ ) - { - int dtype = list[i]->i_type - list[i+1]->i_type; - int dtime = list[i]->i_frame - list[i+1]->i_frame; - int swap = b_dts ? dtype > 0 || ( dtype == 0 && dtime > 0 ) - : dtime > 0; - if( swap ) - { - XCHG( x264_frame_t*, list[i], list[i+1] ); - b_ok = 0; - } - } - } while( !b_ok ); -} - -void x264_weight_scale_plane( x264_t *h, pixel *dst, int i_dst_stride, pixel *src, int i_src_stride, - int i_width, int i_height, x264_weight_t *w ) +void x264_weight_scale_plane( x264_t *h, pixel *dst, intptr_t i_dst_stride, pixel *src, intptr_t i_src_stride, + int i_width, int i_height, x264_weight_t *w ) { /* Weight horizontal strips of height 16. This was found to be the optimal height * in terms of the cache loads. */ while( i_height > 0 ) { - for( int x = 0; x < i_width; x += 16 ) + int x; + for( x = 0; x < i_width-8; x += 16 ) w->weightfn[16>>2]( dst+x, i_dst_stride, src+x, i_src_stride, w, X264_MIN( i_height, 16 ) ); + if( x < i_width ) + w->weightfn[ 8>>2]( dst+x, i_dst_stride, src+x, i_src_stride, w, X264_MIN( i_height, 16 ) ); i_height -= 16; dst += 16 * i_dst_stride; src += 16 * i_src_stride;