avcodec: add D3D12VA hardware accelerated MPEG-2 decoding

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>

7 files changed, 217 insertions, 10 deletions

diff --git a/configure b/configure
index 2fb9dfae3d..e7fd43c9a4 100755
--- a/configure
+++ b/configure
@@ -3132,6 +3132,8 @@ mpeg2_d3d11va_hwaccel_deps="d3d11va"
 mpeg2_d3d11va_hwaccel_select="mpeg2video_decoder"
 mpeg2_d3d11va2_hwaccel_deps="d3d11va"
 mpeg2_d3d11va2_hwaccel_select="mpeg2video_decoder"
+mpeg2_d3d12va_hwaccel_deps="d3d12va"
+mpeg2_d3d12va_hwaccel_select="mpeg2video_decoder"
 mpeg2_dxva2_hwaccel_deps="dxva2"
 mpeg2_dxva2_hwaccel_select="mpeg2video_decoder"
 mpeg2_nvdec_hwaccel_deps="nvdec"
diff --git a/libavcodec/Makefile b/libavcodec/Makefile
index 730dcf373f..889f844031 100644
--- a/libavcodec/Makefile
+++ b/libavcodec/Makefile
@@ -1027,6 +1027,7 @@ OBJS-$(CONFIG_MPEG1_VDPAU_HWACCEL)        += vdpau_mpeg12.o
 OBJS-$(CONFIG_MPEG1_VIDEOTOOLBOX_HWACCEL) += videotoolbox.o
 OBJS-$(CONFIG_MPEG2_D3D11VA_HWACCEL)      += dxva2_mpeg2.o
 OBJS-$(CONFIG_MPEG2_DXVA2_HWACCEL)        += dxva2_mpeg2.o
+OBJS-$(CONFIG_MPEG2_D3D12VA_HWACCEL)      += dxva2_mpeg2.o d3d12va_mpeg2.o
 OBJS-$(CONFIG_MPEG2_NVDEC_HWACCEL)        += nvdec_mpeg12.o
 OBJS-$(CONFIG_MPEG2_QSV_HWACCEL)          += qsvdec.o
 OBJS-$(CONFIG_MPEG2_VAAPI_HWACCEL)        += vaapi_mpeg2.o
diff --git a/libavcodec/d3d12va_mpeg2.c b/libavcodec/d3d12va_mpeg2.c
new file mode 100644
index 0000000000..91bf3f8b75
--- /dev/null
+++ b/libavcodec/d3d12va_mpeg2.c
@@ -0,0 +1,191 @@
+/*
+ * Direct3D12 MPEG-2 HW acceleration
+ *
+ * copyright (c) 2022-2023 Wu Jianhua <toqsxw@outlook.com>
+ *
+ * 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/avassert.h"
+#include "libavutil/hwcontext_d3d12va_internal.h"
+#include "mpegutils.h"
+#include "mpegvideodec.h"
+#include "d3d12va_decode.h"
+#include "dxva2_internal.h"
+
+#define MAX_SLICES  1024
+#define INVALID_REF 0xffff
+
+#define REF_RESOURCE(index) if (index != INVALID_REF) { \
+    ctx->ref_resources[index] = frames_hwctx->texture_infos[index].texture; \
+}
+
+typedef struct D3D12DecodePictureContext {
+    DXVA_PictureParameters  pp;
+    DXVA_QmatrixData        qm;
+    unsigned                slice_count;
+    DXVA_SliceInfo          slices[MAX_SLICES];
+    const uint8_t          *bitstream;
+    unsigned                bitstream_size;
+} D3D12DecodePictureContext;
+
+static int d3d12va_mpeg2_start_frame(AVCodecContext *avctx, av_unused const uint8_t *buffer,  av_unused uint32_t size)
+{
+    const MpegEncContext      *s       = avctx->priv_data;
+    D3D12VADecodeContext      *ctx     = D3D12VA_DECODE_CONTEXT(avctx);
+    D3D12DecodePictureContext *ctx_pic = s->current_picture_ptr->hwaccel_picture_private;
+    DXVA_QmatrixData          *qm      = &ctx_pic->qm;
+
+    if (!ctx)
+        return -1;
+
+    av_assert0(ctx_pic);
+
+    ctx->used_mask = 0;
+
+    ff_dxva2_mpeg2_fill_picture_parameters(avctx, (AVDXVAContext *)ctx, &ctx_pic->pp);
+    ff_dxva2_mpeg2_fill_quantization_matrices(avctx, (AVDXVAContext *)ctx, &ctx_pic->qm);
+
+    // Post processing operations are not supported in D3D12 Video
+    ctx_pic->pp.wDeblockedPictureIndex = INVALID_REF;
+
+    ctx_pic->bitstream      = NULL;
+    ctx_pic->bitstream_size = 0;
+    ctx_pic->slice_count    = 0;
+
+    return 0;
+}
+
+static int d3d12va_mpeg2_decode_slice(AVCodecContext *avctx, const uint8_t *buffer, uint32_t size)
+{
+    const MpegEncContext      *s       = avctx->priv_data;
+    D3D12DecodePictureContext *ctx_pic = s->current_picture_ptr->hwaccel_picture_private;
+
+    int is_field = s->picture_structure != PICT_FRAME;
+
+    if (ctx_pic->slice_count >= MAX_SLICES) {
+        return AVERROR(ERANGE);
+    }
+
+    if (!ctx_pic->bitstream)
+        ctx_pic->bitstream = buffer;
+    ctx_pic->bitstream_size += size;
+
+    ff_dxva2_mpeg2_fill_slice(avctx, &ctx_pic->slices[ctx_pic->slice_count++],
+                              buffer - ctx_pic->bitstream, buffer, size);
+
+    return 0;
+}
+
+static int update_input_arguments(AVCodecContext *avctx, D3D12_VIDEO_DECODE_INPUT_STREAM_ARGUMENTS *input_args, ID3D12Resource *buffer)
+{
+    D3D12VADecodeContext      *ctx          = D3D12VA_DECODE_CONTEXT(avctx);
+    AVHWFramesContext         *frames_ctx   = D3D12VA_FRAMES_CONTEXT(avctx);
+    AVD3D12VAFramesContext    *frames_hwctx = frames_ctx->hwctx;
+    const MpegEncContext      *s            = avctx->priv_data;
+    D3D12DecodePictureContext *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);
+
+    int i;
+    uint8_t *mapped_data = NULL;
+    D3D12_VIDEO_DECODE_FRAME_ARGUMENT *args = &input_args->FrameArguments[input_args->NumFrameArguments++];
+
+    D3D12_RANGE range = {
+        .Begin = 0,
+        .End = ctx_pic->bitstream_size,
+    };
+
+    if (FAILED(ID3D12Resource_Map(buffer, 0, &range, &mapped_data))) {
+        av_log(avctx, AV_LOG_ERROR, "Failed to map D3D12 Buffer resource!\n");
+        return AVERROR(EINVAL);
+    }
+
+    for (i = 0; i < ctx_pic->slice_count; i++) {
+        DXVA_SliceInfo *slice = &ctx_pic->slices[i];
+
+        if (i < ctx_pic->slice_count - 1)
+            slice->wNumberMBsInSlice = slice[1].wNumberMBsInSlice - slice[0].wNumberMBsInSlice;
+        else
+            slice->wNumberMBsInSlice = mb_count - slice[0].wNumberMBsInSlice;
+    }
+
+    memcpy(mapped_data, ctx_pic->bitstream, ctx_pic->bitstream_size);
+
+    ID3D12Resource_Unmap(buffer, 0, &range);
+
+    args->Type  = D3D12_VIDEO_DECODE_ARGUMENT_TYPE_SLICE_CONTROL;
+    args->Size  = sizeof(DXVA_SliceInfo) * ctx_pic->slice_count;
+    args->pData = ctx_pic->slices;
+
+    input_args->CompressedBitstream = (D3D12_VIDEO_DECODE_COMPRESSED_BITSTREAM){
+        .pBuffer = buffer,
+        .Offset  = 0,
+        .Size    = ctx_pic->bitstream_size,
+    };
+
+    return 0;
+}
+
+static int d3d12va_mpeg2_end_frame(AVCodecContext *avctx)
+{
+    int ret;
+    MpegEncContext            *s       = avctx->priv_data;
+    D3D12VADecodeContext      *ctx     = D3D12VA_DECODE_CONTEXT(avctx);
+    D3D12DecodePictureContext *ctx_pic = s->current_picture_ptr->hwaccel_picture_private;
+
+    if (ctx_pic->slice_count <= 0 || ctx_pic->bitstream_size <= 0)
+        return -1;
+
+    ret = ff_d3d12va_common_end_frame(avctx, s->current_picture_ptr->f, &ctx_pic->pp, sizeof(ctx_pic->pp),
+                                      &ctx_pic->qm, sizeof(ctx_pic->qm), update_input_arguments);
+    if (!ret)
+        ff_mpeg_draw_horiz_band(s, 0, avctx->height);
+
+    return ret;
+}
+
+static int d3d12va_mpeg2_decode_init(AVCodecContext *avctx)
+{
+    const MpegEncContext       *s      = avctx->priv_data;
+    D3D12VADecodeContext      *ctx     = D3D12VA_DECODE_CONTEXT(avctx);
+
+    ctx->cfg.DecodeProfile = D3D12_VIDEO_DECODE_PROFILE_MPEG2;
+
+    ctx->max_num_ref = 3;
+
+    return ff_d3d12va_decode_init(avctx);
+}
+
+#if CONFIG_MPEG2_D3D12VA_HWACCEL
+const FFHWAccel ff_mpeg2_d3d12va_hwaccel = {
+    .p.name               = "mpeg2_d3d12va",
+    .p.type               = AVMEDIA_TYPE_VIDEO,
+    .p.id                 = AV_CODEC_ID_MPEG2VIDEO,
+    .p.pix_fmt            = AV_PIX_FMT_D3D12,
+    .init                 = d3d12va_mpeg2_decode_init,
+    .uninit               = ff_d3d12va_decode_uninit,
+    .start_frame          = d3d12va_mpeg2_start_frame,
+    .decode_slice         = d3d12va_mpeg2_decode_slice,
+    .end_frame            = d3d12va_mpeg2_end_frame,
+    .frame_params         = ff_d3d12va_common_frame_params,
+    .frame_priv_data_size = sizeof(D3D12DecodePictureContext),
+    .priv_data_size       = sizeof(D3D12VADecodeContext),
+};
+#endif
diff --git a/libavcodec/dxva2_internal.h b/libavcodec/dxva2_internal.h
index 7a30672ecd..46a359aec6 100644
--- a/libavcodec/dxva2_internal.h
+++ b/libavcodec/dxva2_internal.h
@@ -181,4 +181,10 @@ int ff_dxva2_vp9_fill_picture_parameters(const AVCodecContext *avctx, AVDXVACont
 int ff_dxva2_av1_fill_picture_parameters(const AVCodecContext *avctx, AVDXVAContext *ctx, DXVA_PicParams_AV1 *pp);
 #endif
 
+void ff_dxva2_mpeg2_fill_picture_parameters(AVCodecContext *avctx, AVDXVAContext *ctx, DXVA_PictureParameters *pp);
+
+void ff_dxva2_mpeg2_fill_quantization_matrices(AVCodecContext *avctx, AVDXVAContext *ctx, DXVA_QmatrixData *qm);
+
+void ff_dxva2_mpeg2_fill_slice(AVCodecContext *avctx, DXVA_SliceInfo *slice,  unsigned position, const uint8_t *buffer, unsigned size);
+
 #endif /* AVCODEC_DXVA2_INTERNAL_H */
diff --git a/libavcodec/dxva2_mpeg2.c b/libavcodec/dxva2_mpeg2.c
index b739dc1e54..4c9ed1d761 100644
--- a/libavcodec/dxva2_mpeg2.c
+++ b/libavcodec/dxva2_mpeg2.c
@@ -40,16 +40,15 @@ struct dxva2_picture_context {
     unsigned               bitstream_size;
 };
 
-static void fill_picture_parameters(AVCodecContext *avctx,
+void ff_dxva2_mpeg2_fill_picture_parameters(AVCodecContext *avctx,
                                     AVDXVAContext *ctx,
-                                    const struct MpegEncContext *s,
                                     DXVA_PictureParameters *pp)
 {
+    const struct MpegEncContext *s = avctx->priv_data;
     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);
@@ -59,6 +58,7 @@ static void fill_picture_parameters(AVCodecContext *avctx,
         pp->wBackwardRefPictureIndex = ff_dxva2_get_surface_index(avctx, ctx, s->next_picture.f, 0);
     else
         pp->wBackwardRefPictureIndex = 0xffff;
+    pp->wDecodedPictureIndex         = ff_dxva2_get_surface_index(avctx, ctx, current_picture->f, 1);
     pp->wPicWidthInMBminus1          = s->mb_width  - 1;
     pp->wPicHeightInMBminus1         = (s->mb_height >> is_field) - 1;
     pp->bMacroblockWidthMinus1       = 15;
@@ -106,11 +106,11 @@ static void fill_picture_parameters(AVCodecContext *avctx,
     pp->bBitstreamConcealmentMethod  = 0;
 }
 
-static void fill_quantization_matrices(AVCodecContext *avctx,
+void ff_dxva2_mpeg2_fill_quantization_matrices(AVCodecContext *avctx,
                                        AVDXVAContext *ctx,
-                                       const struct MpegEncContext *s,
                                        DXVA_QmatrixData *qm)
 {
+    const struct MpegEncContext *s = avctx->priv_data;
     int i;
     for (i = 0; i < 4; i++)
         qm->bNewQmatrix[i] = 1;
@@ -123,12 +123,12 @@ static void fill_quantization_matrices(AVCodecContext *avctx,
     }
 }
 
-static void fill_slice(AVCodecContext *avctx,
-                       const struct MpegEncContext *s,
+void ff_dxva2_mpeg2_fill_slice(AVCodecContext *avctx,
                        DXVA_SliceInfo *slice,
                        unsigned position,
                        const uint8_t *buffer, unsigned size)
 {
+    const struct MpegEncContext *s = avctx->priv_data;
     int is_field = s->picture_structure != PICT_FRAME;
     GetBitContext gb;
 
@@ -266,8 +266,8 @@ static int dxva2_mpeg2_start_frame(AVCodecContext *avctx,
         return -1;
     assert(ctx_pic);
 
-    fill_picture_parameters(avctx, ctx, s, &ctx_pic->pp);
-    fill_quantization_matrices(avctx, ctx, s, &ctx_pic->qm);
+    ff_dxva2_mpeg2_fill_picture_parameters(avctx, ctx, &ctx_pic->pp);
+    ff_dxva2_mpeg2_fill_quantization_matrices(avctx, ctx, &ctx_pic->qm);
 
     ctx_pic->slice_count    = 0;
     ctx_pic->bitstream_size = 0;
@@ -293,7 +293,7 @@ static int dxva2_mpeg2_decode_slice(AVCodecContext *avctx,
     ctx_pic->bitstream_size += size;
 
     position = buffer - ctx_pic->bitstream;
-    fill_slice(avctx, s, &ctx_pic->slice[ctx_pic->slice_count++], position,
+    ff_dxva2_mpeg2_fill_slice(avctx, &ctx_pic->slice[ctx_pic->slice_count++], position,
                buffer, size);
     return 0;
 }
diff --git a/libavcodec/hwaccels.h b/libavcodec/hwaccels.h
index 1c4825f2fb..927314c5e3 100644
--- a/libavcodec/hwaccels.h
+++ b/libavcodec/hwaccels.h
@@ -54,6 +54,7 @@ extern const struct FFHWAccel ff_mpeg1_vdpau_hwaccel;
 extern const struct FFHWAccel ff_mpeg1_videotoolbox_hwaccel;
 extern const struct FFHWAccel ff_mpeg2_d3d11va_hwaccel;
 extern const struct FFHWAccel ff_mpeg2_d3d11va2_hwaccel;
+extern const struct FFHWAccel ff_mpeg2_d3d12va_hwaccel;
 extern const struct FFHWAccel ff_mpeg2_dxva2_hwaccel;
 extern const struct FFHWAccel ff_mpeg2_nvdec_hwaccel;
 extern const struct FFHWAccel ff_mpeg2_vaapi_hwaccel;
diff --git a/libavcodec/mpeg12dec.c b/libavcodec/mpeg12dec.c
index 020b7b7564..d07eed8744 100644
--- a/libavcodec/mpeg12dec.c
+++ b/libavcodec/mpeg12dec.c
@@ -850,6 +850,9 @@ static const enum AVPixelFormat mpeg2_hwaccel_pixfmt_list_420[] = {
     AV_PIX_FMT_D3D11VA_VLD,
     AV_PIX_FMT_D3D11,
 #endif
+#if CONFIG_MPEG2_D3D12VA_HWACCEL
+    AV_PIX_FMT_D3D12,
+#endif
 #if CONFIG_MPEG2_VAAPI_HWACCEL
     AV_PIX_FMT_VAAPI,
 #endif
@@ -2620,6 +2623,9 @@ const FFCodec ff_mpeg2video_decoder = {
 #if CONFIG_MPEG2_D3D11VA2_HWACCEL
                         HWACCEL_D3D11VA2(mpeg2),
 #endif
+#if CONFIG_MPEG2_D3D12VA_HWACCEL
+                        HWACCEL_D3D12VA(mpeg2),
+#endif
 #if CONFIG_MPEG2_NVDEC_HWACCEL
                         HWACCEL_NVDEC(mpeg2),
 #endif