FFmpeg/libavcodec/dxva2_mpeg2.c
Wu Jianhua 349ce30e4e avcodec: add D3D12VA hardware accelerated H264 decoding
The implementation is based on:
https://learn.microsoft.com/en-us/windows/win32/medfound/direct3d-12-video-overview

With the Direct3D 12 video decoding support, we can render or process
the decoded images by the pixel shaders or compute shaders directly
without the extra copy overhead, which is beneficial especially if you
are trying to render or post-process a 4K or 8K video.

The command below is how to enable d3d12va:
ffmpeg -hwaccel d3d12va -i input.mp4 output.mp4

Signed-off-by: Wu Jianhua <toqsxw@outlook.com>
Signed-off-by: Tong Wu <tong1.wu@intel.com>
2023-12-21 16:15:23 +08:00

368 lines
14 KiB
C

/*
* MPEG-2 HW acceleration.
*
* copyright (c) 2010 Laurent Aimar
*
* 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
* version 2.1 of the License, or (at your option) any later version.
*
* FFmpeg is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public
* License along with FFmpeg; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
#include "config_components.h"
#include "libavutil/log.h"
#include "dxva2_internal.h"
#include "hwaccel_internal.h"
#include "mpegutils.h"
#include "mpegvideodec.h"
#define MAX_SLICES 1024
struct dxva2_picture_context {
DXVA_PictureParameters pp;
DXVA_QmatrixData qm;
unsigned slice_count;
DXVA_SliceInfo slice[MAX_SLICES];
const uint8_t *bitstream;
unsigned bitstream_size;
};
static void fill_picture_parameters(AVCodecContext *avctx,
AVDXVAContext *ctx,
const struct MpegEncContext *s,
DXVA_PictureParameters *pp)
{
const Picture *current_picture = s->current_picture_ptr;
int is_field = s->picture_structure != PICT_FRAME;
memset(pp, 0, sizeof(*pp));
pp->wDecodedPictureIndex = ff_dxva2_get_surface_index(avctx, ctx, current_picture->f, 1);
pp->wDeblockedPictureIndex = 0;
if (s->pict_type != AV_PICTURE_TYPE_I)
pp->wForwardRefPictureIndex = ff_dxva2_get_surface_index(avctx, ctx, s->last_picture.f, 0);
else
pp->wForwardRefPictureIndex = 0xffff;
if (s->pict_type == AV_PICTURE_TYPE_B)
pp->wBackwardRefPictureIndex = ff_dxva2_get_surface_index(avctx, ctx, s->next_picture.f, 0);
else
pp->wBackwardRefPictureIndex = 0xffff;
pp->wPicWidthInMBminus1 = s->mb_width - 1;
pp->wPicHeightInMBminus1 = (s->mb_height >> is_field) - 1;
pp->bMacroblockWidthMinus1 = 15;
pp->bMacroblockHeightMinus1 = 15;
pp->bBlockWidthMinus1 = 7;
pp->bBlockHeightMinus1 = 7;
pp->bBPPminus1 = 7;
pp->bPicStructure = s->picture_structure;
pp->bSecondField = is_field && !s->first_field;
pp->bPicIntra = s->pict_type == AV_PICTURE_TYPE_I;
pp->bPicBackwardPrediction = s->pict_type == AV_PICTURE_TYPE_B;
pp->bBidirectionalAveragingMode = 0;
pp->bMVprecisionAndChromaRelation= 0; /* FIXME */
pp->bChromaFormat = s->chroma_format;
pp->bPicScanFixed = 1;
pp->bPicScanMethod = s->alternate_scan ? 1 : 0;
pp->bPicReadbackRequests = 0;
pp->bRcontrol = 0;
pp->bPicSpatialResid8 = 0;
pp->bPicOverflowBlocks = 0;
pp->bPicExtrapolation = 0;
pp->bPicDeblocked = 0;
pp->bPicDeblockConfined = 0;
pp->bPic4MVallowed = 0;
pp->bPicOBMC = 0;
pp->bPicBinPB = 0;
pp->bMV_RPS = 0;
pp->bReservedBits = 0;
pp->wBitstreamFcodes = (s->mpeg_f_code[0][0] << 12) |
(s->mpeg_f_code[0][1] << 8) |
(s->mpeg_f_code[1][0] << 4) |
(s->mpeg_f_code[1][1] );
pp->wBitstreamPCEelements = (s->intra_dc_precision << 14) |
(s->picture_structure << 12) |
(s->top_field_first << 11) |
(s->frame_pred_frame_dct << 10) |
(s->concealment_motion_vectors << 9) |
(s->q_scale_type << 8) |
(s->intra_vlc_format << 7) |
(s->alternate_scan << 6) |
(s->repeat_first_field << 5) |
(s->chroma_420_type << 4) |
(s->progressive_frame << 3);
pp->bBitstreamConcealmentNeed = 0;
pp->bBitstreamConcealmentMethod = 0;
}
static void fill_quantization_matrices(AVCodecContext *avctx,
AVDXVAContext *ctx,
const struct MpegEncContext *s,
DXVA_QmatrixData *qm)
{
int i;
for (i = 0; i < 4; i++)
qm->bNewQmatrix[i] = 1;
for (i = 0; i < 64; i++) {
int n = s->idsp.idct_permutation[ff_zigzag_direct[i]];
qm->Qmatrix[0][i] = s->intra_matrix[n];
qm->Qmatrix[1][i] = s->inter_matrix[n];
qm->Qmatrix[2][i] = s->chroma_intra_matrix[n];
qm->Qmatrix[3][i] = s->chroma_inter_matrix[n];
}
}
static void fill_slice(AVCodecContext *avctx,
const struct MpegEncContext *s,
DXVA_SliceInfo *slice,
unsigned position,
const uint8_t *buffer, unsigned size)
{
int is_field = s->picture_structure != PICT_FRAME;
GetBitContext gb;
memset(slice, 0, sizeof(*slice));
slice->wHorizontalPosition = s->mb_x;
slice->wVerticalPosition = s->mb_y >> is_field;
slice->dwSliceBitsInBuffer = 8 * size;
slice->dwSliceDataLocation = position;
slice->bStartCodeBitOffset = 0;
slice->bReservedBits = 0;
/* XXX We store the index of the first MB and it will be fixed later */
slice->wNumberMBsInSlice = (s->mb_y >> is_field) * s->mb_width + s->mb_x;
slice->wBadSliceChopping = 0;
init_get_bits(&gb, &buffer[4], 8 * (size - 4));
slice->wQuantizerScaleCode = get_bits(&gb, 5);
skip_1stop_8data_bits(&gb);
slice->wMBbitOffset = 4 * 8 + get_bits_count(&gb);
}
static int commit_bitstream_and_slice_buffer(AVCodecContext *avctx,
DECODER_BUFFER_DESC *bs,
DECODER_BUFFER_DESC *sc)
{
const struct MpegEncContext *s = avctx->priv_data;
AVDXVAContext *ctx = DXVA_CONTEXT(avctx);
struct dxva2_picture_context *ctx_pic =
s->current_picture_ptr->hwaccel_picture_private;
const int is_field = s->picture_structure != PICT_FRAME;
const unsigned mb_count = s->mb_width * (s->mb_height >> is_field);
void *dxva_data_ptr;
uint8_t *dxva_data, *current, *end;
unsigned dxva_size;
unsigned i;
unsigned type;
#if CONFIG_D3D11VA
if (ff_dxva2_is_d3d11(avctx)) {
type = D3D11_VIDEO_DECODER_BUFFER_BITSTREAM;
if (FAILED(ID3D11VideoContext_GetDecoderBuffer(D3D11VA_CONTEXT(ctx)->video_context,
D3D11VA_CONTEXT(ctx)->decoder,
type,
&dxva_size, &dxva_data_ptr)))
return -1;
}
#endif
#if CONFIG_DXVA2
if (avctx->pix_fmt == AV_PIX_FMT_DXVA2_VLD) {
type = DXVA2_BitStreamDateBufferType;
if (FAILED(IDirectXVideoDecoder_GetBuffer(DXVA2_CONTEXT(ctx)->decoder,
type,
&dxva_data_ptr, &dxva_size)))
return -1;
}
#endif
dxva_data = dxva_data_ptr;
current = dxva_data;
end = dxva_data + dxva_size;
for (i = 0; i < ctx_pic->slice_count; i++) {
DXVA_SliceInfo *slice = &ctx_pic->slice[i];
unsigned position = slice->dwSliceDataLocation;
unsigned size = slice->dwSliceBitsInBuffer / 8;
if (size > end - current) {
av_log(avctx, AV_LOG_ERROR, "Failed to build bitstream");
break;
}
slice->dwSliceDataLocation = current - dxva_data;
if (i < ctx_pic->slice_count - 1)
slice->wNumberMBsInSlice =
slice[1].wNumberMBsInSlice - slice[0].wNumberMBsInSlice;
else
slice->wNumberMBsInSlice =
mb_count - slice[0].wNumberMBsInSlice;
memcpy(current, &ctx_pic->bitstream[position], size);
current += size;
}
#if CONFIG_D3D11VA
if (ff_dxva2_is_d3d11(avctx))
if (FAILED(ID3D11VideoContext_ReleaseDecoderBuffer(D3D11VA_CONTEXT(ctx)->video_context, D3D11VA_CONTEXT(ctx)->decoder, type)))
return -1;
#endif
#if CONFIG_DXVA2
if (avctx->pix_fmt == AV_PIX_FMT_DXVA2_VLD)
if (FAILED(IDirectXVideoDecoder_ReleaseBuffer(DXVA2_CONTEXT(ctx)->decoder, type)))
return -1;
#endif
if (i < ctx_pic->slice_count)
return -1;
#if CONFIG_D3D11VA
if (ff_dxva2_is_d3d11(avctx)) {
D3D11_VIDEO_DECODER_BUFFER_DESC *dsc11 = bs;
memset(dsc11, 0, sizeof(*dsc11));
dsc11->BufferType = type;
dsc11->DataSize = current - dxva_data;
dsc11->NumMBsInBuffer = mb_count;
type = D3D11_VIDEO_DECODER_BUFFER_SLICE_CONTROL;
}
#endif
#if CONFIG_DXVA2
if (avctx->pix_fmt == AV_PIX_FMT_DXVA2_VLD) {
DXVA2_DecodeBufferDesc *dsc2 = bs;
memset(dsc2, 0, sizeof(*dsc2));
dsc2->CompressedBufferType = type;
dsc2->DataSize = current - dxva_data;
dsc2->NumMBsInBuffer = mb_count;
type = DXVA2_SliceControlBufferType;
}
#endif
return ff_dxva2_commit_buffer(avctx, ctx, sc,
type,
ctx_pic->slice,
ctx_pic->slice_count * sizeof(*ctx_pic->slice),
mb_count);
}
static int dxva2_mpeg2_start_frame(AVCodecContext *avctx,
av_unused const uint8_t *buffer,
av_unused uint32_t size)
{
const struct MpegEncContext *s = avctx->priv_data;
AVDXVAContext *ctx = DXVA_CONTEXT(avctx);
struct dxva2_picture_context *ctx_pic =
s->current_picture_ptr->hwaccel_picture_private;
if (!DXVA_CONTEXT_VALID(avctx, ctx))
return -1;
assert(ctx_pic);
fill_picture_parameters(avctx, ctx, s, &ctx_pic->pp);
fill_quantization_matrices(avctx, ctx, s, &ctx_pic->qm);
ctx_pic->slice_count = 0;
ctx_pic->bitstream_size = 0;
ctx_pic->bitstream = NULL;
return 0;
}
static int dxva2_mpeg2_decode_slice(AVCodecContext *avctx,
const uint8_t *buffer, uint32_t size)
{
const struct MpegEncContext *s = avctx->priv_data;
struct dxva2_picture_context *ctx_pic =
s->current_picture_ptr->hwaccel_picture_private;
unsigned position;
if (ctx_pic->slice_count >= MAX_SLICES) {
avpriv_request_sample(avctx, "%d slices in dxva2",
ctx_pic->slice_count);
return -1;
}
if (!ctx_pic->bitstream)
ctx_pic->bitstream = buffer;
ctx_pic->bitstream_size += size;
position = buffer - ctx_pic->bitstream;
fill_slice(avctx, s, &ctx_pic->slice[ctx_pic->slice_count++], position,
buffer, size);
return 0;
}
static int dxva2_mpeg2_end_frame(AVCodecContext *avctx)
{
struct MpegEncContext *s = avctx->priv_data;
struct dxva2_picture_context *ctx_pic =
s->current_picture_ptr->hwaccel_picture_private;
int ret;
if (ctx_pic->slice_count <= 0 || ctx_pic->bitstream_size <= 0)
return -1;
ret = ff_dxva2_common_end_frame(avctx, s->current_picture_ptr->f,
&ctx_pic->pp, sizeof(ctx_pic->pp),
&ctx_pic->qm, sizeof(ctx_pic->qm),
commit_bitstream_and_slice_buffer);
if (!ret)
ff_mpeg_draw_horiz_band(s, 0, avctx->height);
return ret;
}
#if CONFIG_MPEG2_DXVA2_HWACCEL
const FFHWAccel ff_mpeg2_dxva2_hwaccel = {
.p.name = "mpeg2_dxva2",
.p.type = AVMEDIA_TYPE_VIDEO,
.p.id = AV_CODEC_ID_MPEG2VIDEO,
.p.pix_fmt = AV_PIX_FMT_DXVA2_VLD,
.init = ff_dxva2_decode_init,
.uninit = ff_dxva2_decode_uninit,
.start_frame = dxva2_mpeg2_start_frame,
.decode_slice = dxva2_mpeg2_decode_slice,
.end_frame = dxva2_mpeg2_end_frame,
.frame_params = ff_dxva2_common_frame_params,
.frame_priv_data_size = sizeof(struct dxva2_picture_context),
.priv_data_size = sizeof(FFDXVASharedContext),
};
#endif
#if CONFIG_MPEG2_D3D11VA_HWACCEL
const FFHWAccel ff_mpeg2_d3d11va_hwaccel = {
.p.name = "mpeg2_d3d11va",
.p.type = AVMEDIA_TYPE_VIDEO,
.p.id = AV_CODEC_ID_MPEG2VIDEO,
.p.pix_fmt = AV_PIX_FMT_D3D11VA_VLD,
.init = ff_dxva2_decode_init,
.uninit = ff_dxva2_decode_uninit,
.start_frame = dxva2_mpeg2_start_frame,
.decode_slice = dxva2_mpeg2_decode_slice,
.end_frame = dxva2_mpeg2_end_frame,
.frame_params = ff_dxva2_common_frame_params,
.frame_priv_data_size = sizeof(struct dxva2_picture_context),
.priv_data_size = sizeof(FFDXVASharedContext),
};
#endif
#if CONFIG_MPEG2_D3D11VA2_HWACCEL
const FFHWAccel ff_mpeg2_d3d11va2_hwaccel = {
.p.name = "mpeg2_d3d11va2",
.p.type = AVMEDIA_TYPE_VIDEO,
.p.id = AV_CODEC_ID_MPEG2VIDEO,
.p.pix_fmt = AV_PIX_FMT_D3D11,
.init = ff_dxva2_decode_init,
.uninit = ff_dxva2_decode_uninit,
.start_frame = dxva2_mpeg2_start_frame,
.decode_slice = dxva2_mpeg2_decode_slice,
.end_frame = dxva2_mpeg2_end_frame,
.frame_params = ff_dxva2_common_frame_params,
.frame_priv_data_size = sizeof(struct dxva2_picture_context),
.priv_data_size = sizeof(FFDXVASharedContext),
};
#endif