[FFmpeg-devel] [PATCH v1 6/7] lavc/vvc_dec: Add hardware decode API

[Andreas Rheinhardt]

fei.w.wang-at-intel.com at ffmpeg.org:
> From: Fei Wang <fei.w.wang at intel.com>
> 
> Signed-off-by: Fei Wang <fei.w.wang at intel.com>
> ---
>  libavcodec/vvc/vvc_refs.c |  6 ++++
>  libavcodec/vvc/vvcdec.c   | 67 +++++++++++++++++++++++++++++++++++----
>  libavcodec/vvc/vvcdec.h   |  5 +++
>  3 files changed, 72 insertions(+), 6 deletions(-)
> 
> diff --git a/libavcodec/vvc/vvc_refs.c b/libavcodec/vvc/vvc_refs.c
> index bf70777550..c9f89a5a0a 100644
> --- a/libavcodec/vvc/vvc_refs.c
> +++ b/libavcodec/vvc/vvc_refs.c
> @@ -25,6 +25,7 @@
>  #include "libavutil/thread.h"
>  #include "libavcodec/refstruct.h"
>  #include "libavcodec/thread.h"
> +#include "libavcodec/decode.h"
>  
>  #include "vvc_refs.h"
>  
> @@ -56,6 +57,7 @@ void ff_vvc_unref_frame(VVCFrameContext *fc, VVCFrame *frame, int flags)
>          ff_refstruct_unref(&frame->rpl_tab);
>  
>          frame->collocated_ref = NULL;
> +        ff_refstruct_unref(&frame->hwaccel_picture_private);
>      }
>  }
>  
> @@ -138,6 +140,10 @@ static VVCFrame *alloc_frame(VVCContext *s, VVCFrameContext *fc)
>          if (!frame->progress)
>              goto fail;
>  
> +        ret = ff_hwaccel_frame_priv_alloc(s->avctx, &frame->hwaccel_picture_private);
> +        if (ret < 0)
> +            goto fail;
> +
>          return frame;
>  fail:
>          ff_vvc_unref_frame(fc, frame, ~0);
> diff --git a/libavcodec/vvc/vvcdec.c b/libavcodec/vvc/vvcdec.c
> index d5704aca25..f2e269ce76 100644
> --- a/libavcodec/vvc/vvcdec.c
> +++ b/libavcodec/vvc/vvcdec.c
> @@ -24,6 +24,8 @@
>  #include "libavcodec/decode.h"
>  #include "libavcodec/profiles.h"
>  #include "libavcodec/refstruct.h"
> +#include "libavcodec/hwconfig.h"
> +#include "libavcodec/hwaccel_internal.h"
>  #include "libavutil/cpu.h"
>  #include "libavutil/thread.h"
>  
> @@ -563,6 +565,8 @@ static int ref_frame(VVCFrame *dst, const VVCFrame *src)
>  
>      ff_refstruct_replace(&dst->rpl_tab, src->rpl_tab);
>      ff_refstruct_replace(&dst->rpl, src->rpl);
> +    ff_refstruct_replace(&dst->hwaccel_picture_private,
> +                          src->hwaccel_picture_private);
>      dst->nb_rpl_elems = src->nb_rpl_elems;
>  
>      dst->poc = src->poc;
> @@ -718,17 +722,41 @@ static int slice_start(SliceContext *sc, VVCContext *s, VVCFrameContext *fc,
>      return 0;
>  }
>  
> +static enum AVPixelFormat get_format(AVCodecContext *avctx, const VVCSPS *sps)
> +{
> +#define HWACCEL_MAX 0
> +
> +    enum AVPixelFormat pix_fmts[HWACCEL_MAX + 2], *fmt = pix_fmts;
> +
> +    switch (sps->pix_fmt) {
> +    case AV_PIX_FMT_YUV420P:
> +        break;
> +    case AV_PIX_FMT_YUV420P10:
> +        break;
> +    }
> +
> +    *fmt++ = sps->pix_fmt;
> +    *fmt = AV_PIX_FMT_NONE;
> +
> +    return ff_get_format(avctx, pix_fmts);
> +}
> +
>  static void export_frame_params(VVCContext *s, const VVCFrameContext *fc)
>  {
>      AVCodecContext *c = s->avctx;
>      const VVCSPS *sps = fc->ps.sps;
>      const VVCPPS *pps = fc->ps.pps;
>  
> -    c->pix_fmt      = sps->pix_fmt;
> -    c->coded_width  = pps->width;
> -    c->coded_height = pps->height;
> -    c->width        = pps->width  - ((pps->r->pps_conf_win_left_offset + pps->r->pps_conf_win_right_offset) << sps->hshift[CHROMA]);
> -    c->height       = pps->height - ((pps->r->pps_conf_win_top_offset + pps->r->pps_conf_win_bottom_offset) << sps->vshift[CHROMA]);
> +    // Reset HW config if pix_fmt/w/h change.
> +    if (s->pix_fmt != sps->pix_fmt || c->coded_width != pps->width || c->coded_height != pps->height) {
> +        c->coded_width  = pps->width;
> +        c->coded_height = pps->height;
> +        c->pix_fmt      = get_format(c, sps);
> +        s->pix_fmt      = sps->pix_fmt;
> +    }
> +
> +    c->width  = pps->width  - ((pps->r->pps_conf_win_left_offset + pps->r->pps_conf_win_right_offset) << sps->hshift[CHROMA]);
> +    c->height = pps->height - ((pps->r->pps_conf_win_top_offset + pps->r->pps_conf_win_bottom_offset) << sps->vshift[CHROMA]);
>  }
>  
>  static int frame_setup(VVCFrameContext *fc, VVCContext *s)
> @@ -771,6 +799,20 @@ static int decode_slice(VVCContext *s, VVCFrameContext *fc, const H2645NAL *nal,
>      ret = slice_init_entry_points(sc, fc, nal, unit);
>      if (ret < 0)
>          return ret;
> +
> +    if (s->avctx->hwaccel) {
> +        if (is_first_slice) {
> +            ret = FF_HW_CALL(s->avctx, start_frame, NULL, 0);
> +            if (ret < 0)
> +                return ret;
> +        }
> +
> +        ret = FF_HW_CALL(s->avctx, decode_slice,
> +                         nal->raw_data, nal->raw_size);
> +        if (ret < 0)
> +            return ret;
> +    }
> +
>      fc->nb_slices++;
>  
>      return 0;
> @@ -885,9 +927,20 @@ static int wait_delayed_frame(VVCContext *s, AVFrame *output, int *got_output)
>  static int submit_frame(VVCContext *s, VVCFrameContext *fc, AVFrame *output, int *got_output)
>  {
>      int ret;
> +
> +    if (s->avctx->hwaccel) {
> +        if (ret = FF_HW_SIMPLE_CALL(s->avctx, end_frame) < 0) {
> +            av_log(s->avctx, AV_LOG_ERROR,
> +                   "Hardware accelerator failed to decode picture\n");
> +            ff_vvc_unref_frame(fc, fc->ref, ~0);
> +            return ret;
> +        }
> +    } else
> +        ff_vvc_frame_submit(s, fc);
> +
>      s->nb_frames++;
>      s->nb_delayed++;
> -    ff_vvc_frame_submit(s, fc);
> +
>      if (s->nb_delayed >= s->nb_fcs) {
>          if ((ret = wait_delayed_frame(s, output, got_output)) < 0)
>              return ret;
> @@ -1027,6 +1080,8 @@ static av_cold int vvc_decode_init(AVCodecContext *avctx)
>      GDR_SET_RECOVERED(s);
>      ff_thread_once(&init_static_once, init_default_scale_m);
>  
> +    s->pix_fmt = AV_PIX_FMT_NONE;
> +
>      return 0;
>  }
>  
> diff --git a/libavcodec/vvc/vvcdec.h b/libavcodec/vvc/vvcdec.h
> index aa3d715524..009d57424e 100644
> --- a/libavcodec/vvc/vvcdec.h
> +++ b/libavcodec/vvc/vvcdec.h
> @@ -78,6 +78,9 @@ typedef struct VVCFrame {
>       * A combination of VVC_FRAME_FLAG_*
>       */
>      uint8_t flags;
> +
> +    AVBufferRef *hwaccel_priv_buf;

Seems unused (hwaccel_picture_private uses RefStruct nowadays; no other
decoder uses the AVBuffer API for this any more).

> +    void *hwaccel_picture_private; ///< hardware accelerator private data
>  } VVCFrame;
>  
>  typedef struct SliceContext {
> @@ -224,6 +227,8 @@ typedef struct VVCContext {
>  
>      uint64_t nb_frames;     ///< processed frames
>      int nb_delayed;         ///< delayed frames
> +
> +    enum AVPixelFormat pix_fmt; ///< pix format of current frame
>  }  VVCContext ;
>  
>  #endif /* AVCODEC_VVC_VVCDEC_H */