[FFmpeg-devel,v7,1/9] libavutil: add hwcontext_d3d12va and AV_PIX_FMT_D3D12

From: Wu Jianhua <toqsxw@outlook.com>

Signed-off-by: Wu Jianhua <toqsxw@outlook.com>
Signed-off-by: Tong Wu <tong1.wu@intel.com>
---
 configure                              |   5 +
 doc/APIchanges                         |   7 +
 libavutil/Makefile                     |   3 +
 libavutil/hwcontext.c                  |   4 +
 libavutil/hwcontext.h                  |   1 +
 libavutil/hwcontext_d3d12va.c          | 693 +++++++++++++++++++++++++
 libavutil/hwcontext_d3d12va.h          | 148 ++++++
 libavutil/hwcontext_d3d12va_internal.h |  59 +++
 libavutil/hwcontext_internal.h         |   1 +
 libavutil/pixdesc.c                    |   4 +
 libavutil/pixfmt.h                     |   7 +
 libavutil/tests/hwdevice.c             |   2 +
 libavutil/version.h                    |   2 +-
 13 files changed, 935 insertions(+), 1 deletion(-)
 create mode 100644 libavutil/hwcontext_d3d12va.c
 create mode 100644 libavutil/hwcontext_d3d12va.h
 create mode 100644 libavutil/hwcontext_d3d12va_internal.h

V7 removed av_d3d12_va_wait_idle from public API

Thanks,
Tong

>-----Original Message-----
>From: Wu, Tong1 <tong1.wu@intel.com>
>Sent: Tuesday, August 29, 2023 12:10 PM
>To: ffmpeg-devel@ffmpeg.org
>Cc: Wu Jianhua <toqsxw@outlook.com>; Wu, Tong1 <tong1.wu@intel.com>
>Subject: [PATCH v7 1/9] libavutil: add hwcontext_d3d12va and
>AV_PIX_FMT_D3D12
>
>From: Wu Jianhua <toqsxw@outlook.com>
>
>Signed-off-by: Wu Jianhua <toqsxw@outlook.com>
>Signed-off-by: Tong Wu <tong1.wu@intel.com>
>---
> configure                              |   5 +
> doc/APIchanges                         |   7 +
> libavutil/Makefile                     |   3 +
> libavutil/hwcontext.c                  |   4 +
> libavutil/hwcontext.h                  |   1 +
> libavutil/hwcontext_d3d12va.c          | 693 +++++++++++++++++++++++++
> libavutil/hwcontext_d3d12va.h          | 148 ++++++
> libavutil/hwcontext_d3d12va_internal.h |  59 +++
> libavutil/hwcontext_internal.h         |   1 +
> libavutil/pixdesc.c                    |   4 +
> libavutil/pixfmt.h                     |   7 +
> libavutil/tests/hwdevice.c             |   2 +
> libavutil/version.h                    |   2 +-
> 13 files changed, 935 insertions(+), 1 deletion(-)
> create mode 100644 libavutil/hwcontext_d3d12va.c
> create mode 100644 libavutil/hwcontext_d3d12va.h
> create mode 100644 libavutil/hwcontext_d3d12va_internal.h
>
>diff --git a/configure b/configure
>index bd7f7697c8..c604302017 100755
>--- a/configure
>+++ b/configure
>@@ -338,6 +338,7 @@ External library support:
>   --disable-cuda-llvm      disable CUDA compilation using clang [autodetect]
>   --disable-cuvid          disable Nvidia CUVID support [autodetect]
>   --disable-d3d11va        disable Microsoft Direct3D 11 video acceleration code
>[autodetect]
>+  --disable-d3d12va        disable Microsoft Direct3D 12 video acceleration code
>[autodetect]
>   --disable-dxva2          disable Microsoft DirectX 9 video acceleration code
>[autodetect]
>   --disable-ffnvcodec      disable dynamically linked Nvidia code [autodetect]
>   --enable-libdrm          enable DRM code (Linux) [no]
>@@ -1926,6 +1927,7 @@ HWACCEL_AUTODETECT_LIBRARY_LIST="
>     cuda_llvm
>     cuvid
>     d3d11va
>+    d3d12va
>     dxva2
>     ffnvcodec
>     nvdec
>@@ -3054,6 +3056,7 @@ crystalhd_deps="libcrystalhd_libcrystalhd_if_h"
> cuda_deps="ffnvcodec"
> cuvid_deps="ffnvcodec"
> d3d11va_deps="dxva_h ID3D11VideoDecoder ID3D11VideoContext"
>+d3d12va_deps="dxva_h ID3D12Device ID3D12VideoDecoder"
> dxva2_deps="dxva2api_h DXVA2_ConfigPictureDecode ole32 user32"
> ffnvcodec_deps_any="libdl LoadLibrary"
> mediacodec_deps="android"
>@@ -6547,6 +6550,8 @@ check_type "windows.h dxgi1_2.h" "IDXGIOutput1"
> check_type "windows.h dxgi1_5.h" "IDXGIOutput5"
> check_type "windows.h d3d11.h" "ID3D11VideoDecoder"
> check_type "windows.h d3d11.h" "ID3D11VideoContext"
>+check_type "windows.h d3d12.h" "ID3D12Device"
>+check_type "windows.h d3d12video.h" "ID3D12VideoDecoder"
> check_type "windows.h" "DPI_AWARENESS_CONTEXT" -
>D_WIN32_WINNT=0x0A00
> check_type "d3d9.h dxva2api.h" DXVA2_ConfigPictureDecode -
>D_WIN32_WINNT=0x0602
> check_func_headers mfapi.h MFCreateAlignedMemoryBuffer -lmfplat
>diff --git a/doc/APIchanges b/doc/APIchanges
>index ad1efe708d..5004a11267 100644
>--- a/doc/APIchanges
>+++ b/doc/APIchanges
>@@ -2,6 +2,13 @@ The last version increases of all libraries were on 2023-02-
>09
>
> API changes, most recent first:
>
>+2023-07-xx - xxxxxxxxxx - lavu 58.18.100 - pixfmt.h hwcontext.h
>hwcontext_d3d12va.h
>+  Add AV_HWDEVICE_TYPE_D3D12VA and AV_PIX_FMT_D3D12.
>+  Add AVD3D12VADeviceContext, AVD3D12VASyncContext, AVD3D12VAFrame
>and
>+  AVD3D12VAFramesContext.
>+  Add av_d3d12va_map_sw_to_hw_format, av_d3d12va_sync_context_alloc,
>+  av_d3d12va_sync_context_free.
>+
> 2023-08-18 - xxxxxxxxxx - lavu 58.17.100 - channel_layout.h
>   All AV_CHANNEL_LAYOUT_* macros are now compatible with C++ 17 and
>older.
>
>diff --git a/libavutil/Makefile b/libavutil/Makefile
>index 7828c94dc5..db318534eb 100644
>--- a/libavutil/Makefile
>+++ b/libavutil/Makefile
>@@ -41,6 +41,7 @@ HEADERS = adler32.h                                                     \
>           hwcontext.h                                                   \
>           hwcontext_cuda.h                                              \
>           hwcontext_d3d11va.h                                           \
>+          hwcontext_d3d12va.h                                           \
>           hwcontext_drm.h                                               \
>           hwcontext_dxva2.h                                             \
>           hwcontext_qsv.h                                               \
>@@ -188,6 +189,7 @@ OBJS = adler32.o                                                        \
>
> OBJS-$(CONFIG_CUDA)                     += hwcontext_cuda.o
> OBJS-$(CONFIG_D3D11VA)                  += hwcontext_d3d11va.o
>+OBJS-$(CONFIG_D3D12VA)                  += hwcontext_d3d12va.o
> OBJS-$(CONFIG_DXVA2)                    += hwcontext_dxva2.o
> OBJS-$(CONFIG_LIBDRM)                   += hwcontext_drm.o
> OBJS-$(CONFIG_MACOS_KPERF)              += macos_kperf.o
>@@ -211,6 +213,7 @@ SKIPHEADERS-$(HAVE_CUDA_H)             +=
>hwcontext_cuda.h
> SKIPHEADERS-$(CONFIG_CUDA)             += hwcontext_cuda_internal.h     \
>                                           cuda_check.h
> SKIPHEADERS-$(CONFIG_D3D11VA)          += hwcontext_d3d11va.h
>+SKIPHEADERS-$(CONFIG_D3D12VA)          += hwcontext_d3d12va.h
> SKIPHEADERS-$(CONFIG_DXVA2)            += hwcontext_dxva2.h
> SKIPHEADERS-$(CONFIG_QSV)              += hwcontext_qsv.h
> SKIPHEADERS-$(CONFIG_OPENCL)           += hwcontext_opencl.h
>diff --git a/libavutil/hwcontext.c b/libavutil/hwcontext.c
>index 3396598269..04070bc3c3 100644
>--- a/libavutil/hwcontext.c
>+++ b/libavutil/hwcontext.c
>@@ -36,6 +36,9 @@ static const HWContextType * const hw_table[] = {
> #if CONFIG_D3D11VA
>     &ff_hwcontext_type_d3d11va,
> #endif
>+#if CONFIG_D3D12VA
>+    &ff_hwcontext_type_d3d12va,
>+#endif
> #if CONFIG_LIBDRM
>     &ff_hwcontext_type_drm,
> #endif
>@@ -71,6 +74,7 @@ static const char *const hw_type_names[] = {
>     [AV_HWDEVICE_TYPE_DRM]    = "drm",
>     [AV_HWDEVICE_TYPE_DXVA2]  = "dxva2",
>     [AV_HWDEVICE_TYPE_D3D11VA] = "d3d11va",
>+    [AV_HWDEVICE_TYPE_D3D12VA] = "d3d12va",
>     [AV_HWDEVICE_TYPE_OPENCL] = "opencl",
>     [AV_HWDEVICE_TYPE_QSV]    = "qsv",
>     [AV_HWDEVICE_TYPE_VAAPI]  = "vaapi",
>diff --git a/libavutil/hwcontext.h b/libavutil/hwcontext.h
>index 7ff08c8608..2b33721a97 100644
>--- a/libavutil/hwcontext.h
>+++ b/libavutil/hwcontext.h
>@@ -37,6 +37,7 @@ enum AVHWDeviceType {
>     AV_HWDEVICE_TYPE_OPENCL,
>     AV_HWDEVICE_TYPE_MEDIACODEC,
>     AV_HWDEVICE_TYPE_VULKAN,
>+    AV_HWDEVICE_TYPE_D3D12VA,
> };
>
> typedef struct AVHWDeviceInternal AVHWDeviceInternal;
>diff --git a/libavutil/hwcontext_d3d12va.c b/libavutil/hwcontext_d3d12va.c
>new file mode 100644
>index 0000000000..ef0b3572f4
>--- /dev/null
>+++ b/libavutil/hwcontext_d3d12va.c
>@@ -0,0 +1,693 @@
>+/*
>+ * Direct3D 12 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.h"
>+#include "common.h"
>+#include "hwcontext.h"
>+#include "hwcontext_internal.h"
>+#include "hwcontext_d3d12va_internal.h"
>+#include "hwcontext_d3d12va.h"
>+#include "imgutils.h"
>+#include "pixdesc.h"
>+#include "pixfmt.h"
>+#include "thread.h"
>+#include "compat/w32dlfcn.h"
>+#include <dxgi1_3.h>
>+
>+typedef HRESULT(WINAPI *PFN_CREATE_DXGI_FACTORY2)(UINT Flags, REFIID
>riid, void **ppFactory);
>+
>+typedef struct D3D12VAFramesContext {
>+    ID3D12Resource            *staging_buffer;
>+    ID3D12CommandQueue        *command_queue;
>+    ID3D12CommandAllocator    *command_allocator;
>+    ID3D12GraphicsCommandList *command_list;
>+    AVD3D12VASyncContext      *sync_ctx;
>+    int                        nb_surfaces;
>+    int                        nb_surfaces_used;
>+    DXGI_FORMAT                format;
>+    UINT                       luma_component_size;
>+} D3D12VAFramesContext;
>+
>+typedef struct D3D12VADevicePriv {
>+    HANDLE                        d3d12lib;
>+    HANDLE                        dxgilib;
>+    PFN_CREATE_DXGI_FACTORY2      create_dxgi_factory2;
>+    PFN_D3D12_CREATE_DEVICE       create_device;
>+    PFN_D3D12_GET_DEBUG_INTERFACE get_debug_interface;
>+} D3D12VADevicePriv;
>+
>+static const struct {
>+    DXGI_FORMAT d3d_format;
>+    enum AVPixelFormat pix_fmt;
>+} supported_formats[] = {
>+    { DXGI_FORMAT_NV12, AV_PIX_FMT_NV12 },
>+    { DXGI_FORMAT_P010, AV_PIX_FMT_P010 },
>+};
>+
>+DXGI_FORMAT av_d3d12va_map_sw_to_hw_format(enum AVPixelFormat
>pix_fmt)
>+{
>+    switch (pix_fmt) {
>+    case AV_PIX_FMT_NV12:return DXGI_FORMAT_NV12;
>+    case AV_PIX_FMT_P010:return DXGI_FORMAT_P010;
>+    default:             return DXGI_FORMAT_UNKNOWN;
>+    }
>+}
>+
>+static int d3d12va_wait_idle(AVD3D12VASyncContext *sync_ctx)
>+{
>+    uint64_t completion = ID3D12Fence_GetCompletedValue(sync_ctx->fence);
>+    if (completion < sync_ctx->fence_value) {
>+        if (FAILED(ID3D12Fence_SetEventOnCompletion(sync_ctx->fence,
>sync_ctx->fence_value, sync_ctx->event)))
>+            return AVERROR(EINVAL);
>+
>+        WaitForSingleObjectEx(sync_ctx->event, INFINITE, FALSE);
>+    }
>+
>+    return 0;
>+}
>+
>+static inline int d3d12va_wait_queue_idle(AVD3D12VASyncContext *sync_ctx,
>ID3D12CommandQueue *command_queue)
>+{
>+    DX_CHECK(ID3D12CommandQueue_Signal(command_queue, sync_ctx-
>>fence, ++sync_ctx->fence_value));
>+    return d3d12va_wait_idle(sync_ctx);
>+
>+fail:
>+    return AVERROR(EINVAL);
>+}
>+
>+int av_d3d12va_sync_context_alloc(AVD3D12VADeviceContext *ctx,
>AVD3D12VASyncContext **psync_ctx)
>+{
>+    AVD3D12VASyncContext *sync_ctx;
>+
>+    sync_ctx = av_mallocz(sizeof(AVD3D12VASyncContext));
>+    if (!sync_ctx)
>+        return AVERROR(ENOMEM);
>+
>+    DX_CHECK(ID3D12Device_CreateFence(ctx->device, sync_ctx->fence_value,
>D3D12_FENCE_FLAG_NONE, &IID_ID3D12Fence, &sync_ctx->fence));
>+
>+    sync_ctx->event = CreateEvent(NULL, FALSE, FALSE, NULL);
>+    if (!sync_ctx->event)
>+        goto fail;
>+
>+    *psync_ctx = sync_ctx;
>+
>+    return 0;
>+
>+fail:
>+    D3D12_OBJECT_RELEASE(sync_ctx->fence);
>+    av_freep(&sync_ctx);
>+    return AVERROR(EINVAL);
>+}
>+
>+void av_d3d12va_sync_context_free(AVD3D12VASyncContext **psync_ctx)
>+{
>+    AVD3D12VASyncContext *sync_ctx;
>+
>+    if (!psync_ctx || !*psync_ctx)
>+        return;
>+
>+    sync_ctx = *psync_ctx;
>+
>+    d3d12va_wait_idle(sync_ctx);
>+
>+    D3D12_OBJECT_RELEASE(sync_ctx->fence);
>+
>+    if (sync_ctx->event)
>+        CloseHandle(sync_ctx->event);
>+
>+    av_freep(psync_ctx);
>+}
>+
>+static inline int create_resource(ID3D12Device *device, const
>D3D12_RESOURCE_DESC *desc, D3D12_RESOURCE_STATES states,
>ID3D12Resource **ppResource, int is_read_back)
>+{
>+    D3D12_HEAP_PROPERTIES props = {
>+        .Type                 = is_read_back ? D3D12_HEAP_TYPE_READBACK :
>D3D12_HEAP_TYPE_DEFAULT,
>+        .CPUPageProperty      = D3D12_CPU_PAGE_PROPERTY_UNKNOWN,
>+        .MemoryPoolPreference = D3D12_MEMORY_POOL_UNKNOWN,
>+        .CreationNodeMask     = 0,
>+        .VisibleNodeMask      = 0,
>+    };
>+
>+    if (FAILED(ID3D12Device_CreateCommittedResource(device, &props,
>D3D12_HEAP_FLAG_NONE, desc,
>+        states, NULL, &IID_ID3D12Resource, ppResource)))
>+        return AVERROR(EINVAL);
>+
>+    return 0;
>+}
>+
>+static int d3d12va_create_staging_buffer_resource(AVHWFramesContext *ctx)
>+{
>+    AVD3D12VADeviceContext *device_hwctx = ctx->device_ctx->hwctx;
>+    D3D12VAFramesContext   *s            = ctx->internal->priv;
>+
>+    D3D12_RESOURCE_DESC desc = {
>+        .Dimension          = D3D12_RESOURCE_DIMENSION_BUFFER,
>+        .Alignment          = 0,
>+        .Width              = 0,
>+        .Height             = 1,
>+        .DepthOrArraySize   = 1,
>+        .MipLevels          = 1,
>+        .Format             = DXGI_FORMAT_UNKNOWN,
>+        .SampleDesc         = { .Count = 1, .Quality = 0 },
>+        .Layout             = D3D12_TEXTURE_LAYOUT_ROW_MAJOR,
>+        .Flags              = D3D12_RESOURCE_FLAG_NONE,
>+    };
>+
>+    s->luma_component_size = FFALIGN(ctx->width * (s->format ==
>DXGI_FORMAT_P010 ? 2 : 1), D3D12_TEXTURE_DATA_PITCH_ALIGNMENT) *
>ctx->height;
>+    desc.Width = s->luma_component_size + (s->luma_component_size >> 1);
>+
>+    return create_resource(device_hwctx->device, &desc,
>D3D12_RESOURCE_STATE_COPY_DEST, &s->staging_buffer, 1);
>+}
>+
>+static int d3d12va_create_helper_objects(AVHWFramesContext *ctx)
>+{
>+    AVD3D12VADeviceContext *device_hwctx = ctx->device_ctx->hwctx;
>+    D3D12VAFramesContext   *s            = ctx->internal->priv;
>+
>+    D3D12_COMMAND_QUEUE_DESC queue_desc = {
>+        .Type     = D3D12_COMMAND_LIST_TYPE_COPY,
>+        .Priority = 0,
>+        .NodeMask = 0,
>+    };
>+
>+    int ret = d3d12va_create_staging_buffer_resource(ctx);
>+    if (ret < 0)
>+        return ret;
>+
>+    ret = av_d3d12va_sync_context_alloc(device_hwctx, &s->sync_ctx);
>+    if (ret < 0)
>+        return ret;
>+
>+    DX_CHECK(ID3D12Device_CreateCommandQueue(device_hwctx->device,
>&queue_desc,
>+        &IID_ID3D12CommandQueue, &s->command_queue));
>+
>+    DX_CHECK(ID3D12Device_CreateCommandAllocator(device_hwctx->device,
>queue_desc.Type,
>+        &IID_ID3D12CommandAllocator, &s->command_allocator));
>+
>+    DX_CHECK(ID3D12Device_CreateCommandList(device_hwctx->device, 0,
>queue_desc.Type,
>+        s->command_allocator, NULL, &IID_ID3D12GraphicsCommandList, &s-
>>command_list));
>+
>+    DX_CHECK(ID3D12GraphicsCommandList_Close(s->command_list));
>+
>+    ID3D12CommandQueue_ExecuteCommandLists(s->command_queue, 1,
>(ID3D12CommandList **)&s->command_list);
>+
>+    return d3d12va_wait_queue_idle(s->sync_ctx, s->command_queue);
>+
>+fail:
>+    return AVERROR(EINVAL);
>+}
>+
>+static void d3d12va_frames_uninit(AVHWFramesContext *ctx)
>+{
>+    AVD3D12VAFramesContext *frames_hwctx = ctx->hwctx;
>+    D3D12VAFramesContext   *s            = ctx->internal->priv;
>+
>+    av_d3d12va_sync_context_free(&s->sync_ctx);
>+
>+    D3D12_OBJECT_RELEASE(s->staging_buffer);
>+    D3D12_OBJECT_RELEASE(s->command_allocator);
>+    D3D12_OBJECT_RELEASE(s->command_list);
>+    D3D12_OBJECT_RELEASE(s->command_queue);
>+
>+    av_freep(&frames_hwctx->texture_infos);
>+}
>+
>+static int d3d12va_frames_get_constraints(AVHWDeviceContext *ctx, const
>void *hwconfig, AVHWFramesConstraints *constraints)
>+{
>+    HRESULT hr;
>+    int nb_sw_formats = 0;
>+    AVD3D12VADeviceContext *device_hwctx = ctx->hwctx;
>+
>+    constraints->valid_sw_formats =
>av_malloc_array(FF_ARRAY_ELEMS(supported_formats) + 1,
>+                                                    sizeof(*constraints->valid_sw_formats));
>+    if (!constraints->valid_sw_formats)
>+        return AVERROR(ENOMEM);
>+
>+    for (int i = 0; i < FF_ARRAY_ELEMS(supported_formats); i++) {
>+        D3D12_FEATURE_DATA_FORMAT_SUPPORT format_support =
>{ supported_formats[i].d3d_format };
>+        hr = ID3D12Device_CheckFeatureSupport(device_hwctx->device,
>D3D12_FEATURE_FORMAT_SUPPORT, &format_support,
>sizeof(format_support));
>+        if (SUCCEEDED(hr) && (format_support.Support1 &
>D3D12_FORMAT_SUPPORT1_TEXTURE2D))
>+            constraints->valid_sw_formats[nb_sw_formats++] =
>supported_formats[i].pix_fmt;
>+    }
>+    constraints->valid_sw_formats[nb_sw_formats] = AV_PIX_FMT_NONE;
>+
>+    constraints->valid_hw_formats = av_malloc_array(2, sizeof(*constraints-
>>valid_hw_formats));
>+    if (!constraints->valid_hw_formats)
>+        return AVERROR(ENOMEM);
>+
>+    constraints->valid_hw_formats[0] = AV_PIX_FMT_D3D12;
>+    constraints->valid_hw_formats[1] = AV_PIX_FMT_NONE;
>+
>+    return 0;
>+}
>+
>+static void free_texture(void *opaque, uint8_t *data)
>+{
>+    AVD3D12VAFrame *frame = (AVD3D12VAFrame *)data;
>+
>+    if (frame->sync_ctx)
>+        av_d3d12va_sync_context_free(&frame->sync_ctx);
>+
>+    D3D12_OBJECT_RELEASE(frame->texture);
>+    av_freep(&data);
>+}
>+
>+static AVBufferRef *d3d12va_pool_alloc(void *opaque, size_t size)
>+{
>+    AVHWFramesContext      *ctx          = (AVHWFramesContext *)opaque;
>+    D3D12VAFramesContext   *s            = ctx->internal->priv;
>+    AVD3D12VAFramesContext *hwctx        = ctx->hwctx;
>+    AVD3D12VADeviceContext *device_hwctx = ctx->device_ctx->hwctx;
>+
>+    int ret;
>+    AVBufferRef *buf;
>+    AVD3D12VAFrame *frame;
>+
>+    D3D12_RESOURCE_DESC desc = {
>+        .Dimension        = D3D12_RESOURCE_DIMENSION_TEXTURE2D,
>+        .Alignment        = 0,
>+        .Width            = ctx->width,
>+        .Height           = ctx->height,
>+        .DepthOrArraySize = 1,
>+        .MipLevels        = 1,
>+        .Format           = s->format,
>+        .SampleDesc       = {.Count = 1, .Quality = 0 },
>+        .Layout           = D3D12_TEXTURE_LAYOUT_UNKNOWN,
>+        .Flags            = D3D12_RESOURCE_FLAG_NONE,
>+    };
>+
>+    if (s->nb_surfaces_used >= ctx->initial_pool_size) {
>+        av_log(ctx, AV_LOG_ERROR, "Static surface pool size exceeded.\n");
>+        return NULL;
>+    }
>+
>+    frame = av_mallocz(sizeof(AVD3D12VAFrame));
>+    if (!frame)
>+        return NULL;
>+
>+    ret = create_resource(device_hwctx->device, &desc,
>D3D12_RESOURCE_STATE_COMMON, &frame->texture, 0);
>+    if (ret < 0)
>+        goto fail;
>+
>+    ret = av_d3d12va_sync_context_alloc(device_hwctx, &frame->sync_ctx);
>+    if (ret < 0)
>+        goto fail;
>+
>+    frame->index = s->nb_surfaces_used;
>+    hwctx->texture_infos[s->nb_surfaces_used].texture = frame->texture;
>+    hwctx->texture_infos[s->nb_surfaces_used].index = frame->index;
>+    hwctx->texture_infos[s->nb_surfaces_used].sync_ctx = frame->sync_ctx;
>+    s->nb_surfaces_used++;
>+
>+    buf = av_buffer_create((uint8_t *)frame, sizeof(frame), free_texture, NULL,
>0);
>+    if (!buf)
>+        goto fail;
>+
>+    return buf;
>+
>+fail:
>+    free_texture(NULL, (uint8_t *)frame);
>+    return NULL;
>+}
>+
>+static int d3d12va_frames_init(AVHWFramesContext *ctx)
>+{
>+    AVD3D12VAFramesContext *hwctx        = ctx->hwctx;
>+    AVD3D12VADeviceContext *device_hwctx = ctx->device_ctx->hwctx;
>+    D3D12VAFramesContext   *s            = ctx->internal->priv;
>+
>+    int i;
>+
>+    for (i = 0; i < FF_ARRAY_ELEMS(supported_formats); i++) {
>+        if (ctx->sw_format == supported_formats[i].pix_fmt) {
>+            s->format = supported_formats[i].d3d_format;
>+            break;
>+        }
>+    }
>+    if (i == FF_ARRAY_ELEMS(supported_formats)) {
>+        av_log(ctx, AV_LOG_ERROR, "Unsupported pixel format: %s\n",
>+               av_get_pix_fmt_name(ctx->sw_format));
>+        return AVERROR(EINVAL);
>+    }
>+
>+    hwctx->texture_infos = av_realloc_f(NULL, ctx->initial_pool_size,
>sizeof(*hwctx->texture_infos));
>+    if (!hwctx->texture_infos)
>+        return AVERROR(ENOMEM);
>+
>+    memset(hwctx->texture_infos, 0, ctx->initial_pool_size * sizeof(*hwctx-
>>texture_infos));
>+    s->nb_surfaces = ctx->initial_pool_size;
>+
>+    ctx->internal->pool_internal =
>av_buffer_pool_init2(sizeof(AVD3D12VAFrame),
>+        ctx, d3d12va_pool_alloc, NULL);
>+
>+    if (!ctx->internal->pool_internal)
>+        return AVERROR(ENOMEM);
>+
>+    return 0;
>+}
>+
>+static int d3d12va_get_buffer(AVHWFramesContext *ctx, AVFrame *frame)
>+{
>+    int ret;
>+
>+    frame->buf[0] = av_buffer_pool_get(ctx->pool);
>+    if (!frame->buf[0])
>+        return AVERROR(ENOMEM);
>+
>+    ret = av_image_fill_arrays(frame->data, frame->linesize, NULL,
>+        ctx->sw_format, ctx->width, ctx->height,
>D3D12_TEXTURE_DATA_PITCH_ALIGNMENT);
>+    if (ret < 0)
>+        return ret;
>+
>+    frame->data[0] = frame->buf[0]->data;
>+    frame->format  = AV_PIX_FMT_D3D12;
>+    frame->width   = ctx->width;
>+    frame->height  = ctx->height;
>+
>+    return 0;
>+}
>+
>+static int d3d12va_transfer_get_formats(AVHWFramesContext *ctx,
>+                                        enum AVHWFrameTransferDirection dir,
>+                                        enum AVPixelFormat **formats)
>+{
>+    D3D12VAFramesContext *s = ctx->internal->priv;
>+    enum AVPixelFormat *fmts;
>+
>+    fmts = av_malloc_array(2, sizeof(*fmts));
>+    if (!fmts)
>+        return AVERROR(ENOMEM);
>+
>+    fmts[0] = ctx->sw_format;
>+    fmts[1] = AV_PIX_FMT_NONE;
>+
>+    *formats = fmts;
>+
>+    return 0;
>+}
>+
>+static int d3d12va_transfer_data(AVHWFramesContext *ctx, AVFrame *dst,
>+                                 const AVFrame *src)
>+{
>+    AVD3D12VADeviceContext *hwctx        = ctx->device_ctx->hwctx;
>+    AVD3D12VAFramesContext *frames_hwctx = ctx->hwctx;
>+    D3D12VAFramesContext   *s            = ctx->internal->priv;
>+
>+    int ret;
>+    int download = src->format == AV_PIX_FMT_D3D12;
>+    const AVFrame *frame = download ? src : dst;
>+    const AVFrame *other = download ? dst : src;
>+
>+    AVD3D12VAFrame *f = (AVD3D12VAFrame *)frame->data[0];
>+    ID3D12Resource       *texture  = (ID3D12Resource *)      f->texture;
>+    int                   index    = (intptr_t)              f->index;
>+    AVD3D12VASyncContext *sync_ctx = (AVD3D12VASyncContext *)f-
>>sync_ctx;
>+
>+    uint8_t *mapped_data;
>+    uint8_t *data[4];
>+    int linesizes[4];
>+
>+    D3D12_TEXTURE_COPY_LOCATION staging_y_location;
>+    D3D12_TEXTURE_COPY_LOCATION staging_uv_location;
>+
>+    D3D12_TEXTURE_COPY_LOCATION texture_y_location = {
>+        .pResource        = texture,
>+        .Type             = D3D12_TEXTURE_COPY_TYPE_SUBRESOURCE_INDEX,
>+        .SubresourceIndex = 0,
>+    };
>+
>+    D3D12_TEXTURE_COPY_LOCATION texture_uv_location = {
>+        .pResource        = texture,
>+        .Type             = D3D12_TEXTURE_COPY_TYPE_SUBRESOURCE_INDEX,
>+        .SubresourceIndex = 1,
>+    };
>+
>+    D3D12_RESOURCE_BARRIER barrier = {
>+        .Type  = D3D12_RESOURCE_BARRIER_TYPE_TRANSITION,
>+        .Flags = D3D12_RESOURCE_BARRIER_FLAG_NONE,
>+        .Transition = {
>+            .pResource   = texture,
>+            .StateBefore = D3D12_RESOURCE_STATE_COMMON,
>+            .StateAfter  = D3D12_RESOURCE_STATE_COPY_SOURCE,
>+            .Subresource = D3D12_RESOURCE_BARRIER_ALL_SUBRESOURCES,
>+        }
>+    };
>+
>+    s->format = av_d3d12va_map_sw_to_hw_format(ctx->sw_format);
>+
>+    if (frame->hw_frames_ctx->data != (uint8_t *)ctx || other->format != ctx-
>>sw_format)
>+        return AVERROR(EINVAL);
>+
>+    if (!s->command_queue) {
>+        ret = d3d12va_create_helper_objects(ctx);
>+        if (ret < 0)
>+            return ret;
>+    }
>+
>+    for (int i = 0; i < 4; i++)
>+        linesizes[i] = FFALIGN(frame->width * (s->format ==
>DXGI_FORMAT_P010 ? 2 : 1), D3D12_TEXTURE_DATA_PITCH_ALIGNMENT);
>+
>+    staging_y_location = (D3D12_TEXTURE_COPY_LOCATION) {
>+        .pResource = s->staging_buffer,
>+        .Type      = D3D12_TEXTURE_COPY_TYPE_PLACED_FOOTPRINT,
>+        .PlacedFootprint = {
>+            .Offset = 0,
>+            .Footprint = {
>+                .Format   = s->format == DXGI_FORMAT_P010 ?
>DXGI_FORMAT_R16_UNORM : DXGI_FORMAT_R8_UNORM,
>+                .Width    = ctx->width,
>+                .Height   = ctx->height,
>+                .Depth    = 1,
>+                .RowPitch = linesizes[0],
>+            },
>+        },
>+    };
>+
>+    staging_uv_location = (D3D12_TEXTURE_COPY_LOCATION) {
>+        .pResource = s->staging_buffer,
>+        .Type      = D3D12_TEXTURE_COPY_TYPE_PLACED_FOOTPRINT,
>+        .PlacedFootprint = {
>+            .Offset = s->luma_component_size,
>+            .Footprint = {
>+                .Format   = s->format == DXGI_FORMAT_P010 ?
>DXGI_FORMAT_R16G16_UNORM : DXGI_FORMAT_R8G8_UNORM,
>+                .Width    = ctx->width  >> 1,
>+                .Height   = ctx->height >> 1,
>+                .Depth    = 1,
>+                .RowPitch = linesizes[0],
>+            },
>+        },
>+    };
>+
>+    DX_CHECK(ID3D12CommandAllocator_Reset(s->command_allocator));
>+
>+    DX_CHECK(ID3D12GraphicsCommandList_Reset(s->command_list, s-
>>command_allocator, NULL));
>+
>+    if (download) {
>+        ID3D12GraphicsCommandList_ResourceBarrier(s->command_list, 1,
>&barrier);
>+
>+        ID3D12GraphicsCommandList_CopyTextureRegion(s->command_list,
>+            &staging_y_location, 0, 0, 0, &texture_y_location, NULL);
>+
>+        ID3D12GraphicsCommandList_CopyTextureRegion(s->command_list,
>+            &staging_uv_location, 0, 0, 0, &texture_uv_location, NULL);
>+
>+        barrier.Transition.StateBefore = barrier.Transition.StateAfter;
>+        barrier.Transition.StateAfter = D3D12_RESOURCE_STATE_COMMON;
>+        ID3D12GraphicsCommandList_ResourceBarrier(s->command_list, 1,
>&barrier);
>+
>+        DX_CHECK(ID3D12GraphicsCommandList_Close(s->command_list));
>+
>+        if (!hwctx->sync)
>+            DX_CHECK(ID3D12CommandQueue_Wait(s->command_queue,
>sync_ctx->fence, sync_ctx->fence_value));
>+
>+        ID3D12CommandQueue_ExecuteCommandLists(s->command_queue, 1,
>(ID3D12CommandList **)&s->command_list);
>+
>+        ret = d3d12va_wait_queue_idle(s->sync_ctx, s->command_queue);
>+        if (ret)
>+            return ret;
>+
>+        DX_CHECK(ID3D12Resource_Map(s->staging_buffer, 0, NULL,
>&mapped_data));
>+        av_image_fill_pointers(data, ctx->sw_format, ctx->height, mapped_data,
>linesizes);
>+
>+        av_image_copy(dst->data, dst->linesize, data, linesizes,
>+            ctx->sw_format, ctx->width, ctx->height);
>+
>+        ID3D12Resource_Unmap(s->staging_buffer, 0, NULL);
>+    } else {
>+        av_log(ctx, AV_LOG_ERROR, "Transfer data to AV_PIX_FMT_D3D12 is not
>supported yet!\n");
>+        return AVERROR(EINVAL);
>+    }
>+
>+    return 0;
>+
>+fail:
>+    return AVERROR(EINVAL);
>+}
>+
>+static int d3d12va_load_functions(AVHWDeviceContext *hwdev)
>+{
>+    D3D12VADevicePriv *priv = hwdev->internal->priv;
>+
>+#if !HAVE_UWP
>+    priv->d3d12lib = dlopen("d3d12.dll", 0);
>+    priv->dxgilib  = dlopen("dxgi.dll", 0);
>+
>+    if (!priv->d3d12lib || !priv->dxgilib)
>+        goto fail;
>+
>+    priv->create_device = (PFN_D3D12_CREATE_DEVICE)GetProcAddress(priv-
>>d3d12lib, "D3D12CreateDevice");
>+    if (!priv->create_device)
>+        goto fail;
>+
>+    priv->create_dxgi_factory2 =
>(PFN_CREATE_DXGI_FACTORY2)GetProcAddress(priv->dxgilib,
>"CreateDXGIFactory2");
>+    if (!priv->create_dxgi_factory2)
>+        goto fail;
>+
>+    priv->get_debug_interface  =
>(PFN_D3D12_GET_DEBUG_INTERFACE)GetProcAddress(priv->d3d12lib,
>"D3D12GetDebugInterface");
>+#else
>+    priv->create_device        = (PFN_D3D12_CREATE_DEVICE)
>D3D12CreateDevice;
>+    priv->create_dxgi_factory2 = (PFN_CREATE_DXGI_FACTORY2)
>CreateDXGIFactory2;
>+    priv->get_debug_interface  = (PFN_D3D12_GET_DEBUG_INTERFACE)
>D3D12GetDebugInterface;
>+#endif
>+    return 0;
>+
>+fail:
>+    av_log(hwdev, AV_LOG_ERROR, "Failed to load D3D12 library or its
>functions\n");
>+    return AVERROR_UNKNOWN;
>+}
>+
>+static void d3d12va_device_free(AVHWDeviceContext *hwdev)
>+{
>+    AVD3D12VADeviceContext *ctx  = hwdev->hwctx;
>+    D3D12VADevicePriv      *priv = hwdev->internal->priv;
>+
>+    D3D12_OBJECT_RELEASE(ctx->device);
>+
>+    if (priv->d3d12lib)
>+        dlclose(priv->d3d12lib);
>+
>+    if (priv->dxgilib)
>+        dlclose(priv->dxgilib);
>+}
>+
>+static int d3d12va_device_init(AVHWDeviceContext *hwdev)
>+{
>+    AVD3D12VADeviceContext *ctx = hwdev->hwctx;
>+
>+    if (!ctx->video_device)
>+        DX_CHECK(ID3D12Device_QueryInterface(ctx->device,
>&IID_ID3D12VideoDevice, (void **)&ctx->video_device));
>+
>+    return 0;
>+
>+fail:
>+    return AVERROR(EINVAL);
>+}
>+
>+static void d3d12va_device_uninit(AVHWDeviceContext *hwdev)
>+{
>+    AVD3D12VADeviceContext *device_hwctx = hwdev->hwctx;
>+
>+    D3D12_OBJECT_RELEASE(device_hwctx->video_device);
>+}
>+
>+static int d3d12va_device_create(AVHWDeviceContext *hwdev, const char
>*device,
>+                                 AVDictionary *opts, int flags)
>+{
>+    AVD3D12VADeviceContext *ctx  = hwdev->hwctx;
>+    D3D12VADevicePriv      *priv = hwdev->internal->priv;
>+
>+    HRESULT hr;
>+    UINT create_flags = 0;
>+    IDXGIAdapter *pAdapter = NULL;
>+
>+    int ret;
>+    int is_debug = !!av_dict_get(opts, "debug", NULL, 0);
>+    ctx->sync    = !!av_dict_get(opts, "sync",  NULL, 0);
>+
>+    hwdev->free = d3d12va_device_free;
>+
>+    ret = d3d12va_load_functions(hwdev);
>+    if (ret < 0)
>+        return ret;
>+
>+    if (is_debug) {
>+        ID3D12Debug *pDebug;
>+        if (priv->get_debug_interface && SUCCEEDED(priv-
>>get_debug_interface(&IID_ID3D12Debug, &pDebug))) {
>+            create_flags |= DXGI_CREATE_FACTORY_DEBUG;
>+            ID3D12Debug_EnableDebugLayer(pDebug);
>+            D3D12_OBJECT_RELEASE(pDebug);
>+            av_log(hwdev, AV_LOG_INFO, "D3D12 debug layer is enabled!\n");
>+        }
>+    }
>+
>+    if (!ctx->device) {
>+        IDXGIFactory2 *pDXGIFactory = NULL;
>+
>+        hr = priv->create_dxgi_factory2(create_flags, &IID_IDXGIFactory2, (void
>**)&pDXGIFactory);
>+        if (SUCCEEDED(hr)) {
>+            int adapter = device ? atoi(device) : 0;
>+            if (FAILED(IDXGIFactory2_EnumAdapters(pDXGIFactory, adapter,
>&pAdapter)))
>+                pAdapter = NULL;
>+            IDXGIFactory2_Release(pDXGIFactory);
>+        }
>+
>+        if (pAdapter) {
>+            DXGI_ADAPTER_DESC desc;
>+            hr = IDXGIAdapter2_GetDesc(pAdapter, &desc);
>+            if (!FAILED(hr)) {
>+                av_log(ctx, AV_LOG_INFO, "Using device %04x:%04x (%ls).\n",
>+                       desc.VendorId, desc.DeviceId, desc.Description);
>+            }
>+        }
>+
>+        hr = priv->create_device((IUnknown *)pAdapter,
>D3D_FEATURE_LEVEL_12_0, &IID_ID3D12Device, &ctx->device);
>+        D3D12_OBJECT_RELEASE(pAdapter);
>+        if (FAILED(hr)) {
>+            av_log(ctx, AV_LOG_ERROR, "Failed to create Direct 3D 12 device
>(%lx)\n", (long)hr);
>+            return AVERROR_UNKNOWN;
>+        }
>+    }
>+
>+    return 0;
>+}
>+
>+const HWContextType ff_hwcontext_type_d3d12va = {
>+    .type                   = AV_HWDEVICE_TYPE_D3D12VA,
>+    .name                   = "D3D12VA",
>+
>+    .device_hwctx_size      = sizeof(AVD3D12VADeviceContext),
>+    .device_priv_size       = sizeof(D3D12VADevicePriv),
>+    .frames_hwctx_size      = sizeof(AVD3D12VAFramesContext),
>+    .frames_priv_size       = sizeof(D3D12VAFramesContext),
>+
>+    .device_create          = d3d12va_device_create,
>+    .device_init            = d3d12va_device_init,
>+    .device_uninit          = d3d12va_device_uninit,
>+    .frames_get_constraints = d3d12va_frames_get_constraints,
>+    .frames_init            = d3d12va_frames_init,
>+    .frames_uninit          = d3d12va_frames_uninit,
>+    .frames_get_buffer      = d3d12va_get_buffer,
>+    .transfer_get_formats   = d3d12va_transfer_get_formats,
>+    .transfer_data_to       = d3d12va_transfer_data,
>+    .transfer_data_from     = d3d12va_transfer_data,
>+
>+    .pix_fmts               = (const enum AVPixelFormat[]){ AV_PIX_FMT_D3D12,
>AV_PIX_FMT_NONE },
>+};
>diff --git a/libavutil/hwcontext_d3d12va.h b/libavutil/hwcontext_d3d12va.h
>new file mode 100644
>index 0000000000..cb4dc60ef0
>--- /dev/null
>+++ b/libavutil/hwcontext_d3d12va.h
>@@ -0,0 +1,148 @@
>+/*
>+ * Direct3D 12 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
>+ */
>+
>+#ifndef AVUTIL_HWCONTEXT_D3D12VA_H
>+#define AVUTIL_HWCONTEXT_D3D12VA_H
>+
>+/**
>+ * @file
>+ * An API-specific header for AV_HWDEVICE_TYPE_D3D12VA.
>+ *
>+ * This API does not support dynamic frame pools. AVHWFramesContext.pool
>must
>+ * contain AVBufferRefs whose data pointer points to an
>AVD3D12FrameDescriptor struct.
>+ */
>+#include <stdint.h>
>+#include <initguid.h>
>+#include <d3d12.h>
>+#include <d3d12video.h>
>+
>+/**
>+ * @brief This struct is allocated as AVHWDeviceContext.hwctx
>+ *
>+ */
>+typedef struct AVD3D12VADeviceContext {
>+    /**
>+     * Device used for objects creation and access. This can also be
>+     * used to set the libavcodec decoding device.
>+     *
>+     * Can be set by the user. This is the only mandatory field - the other
>+     * device context fields are set from this and are available for convenience.
>+     *
>+     * Deallocating the AVHWDeviceContext will always release this interface,
>+     * and it does not matter whether it was user-allocated.
>+     */
>+    ID3D12Device *device;
>+
>+    /**
>+     * If unset, this will be set from the device field on init.
>+     *
>+     * Deallocating the AVHWDeviceContext will always release this interface,
>+     * and it does not matter whether it was user-allocated.
>+     */
>+    ID3D12VideoDevice *video_device;
>+
>+    /**
>+     * Specifed by sync=1 when init d3d12va
>+     *
>+     * Execute commands as sync mode
>+     */
>+    int sync;
>+} AVD3D12VADeviceContext;
>+
>+/**
>+ * @brief This struct is used to sync d3d12 execution
>+ *
>+ */
>+typedef struct AVD3D12VASyncContext {
>+    /**
>+     * D3D12 fence object
>+     */
>+    ID3D12Fence *fence;
>+
>+    /**
>+     * A handle to the event object
>+     */
>+    HANDLE event;
>+
>+    /**
>+     * The fence value used for sync
>+     */
>+    uint64_t fence_value;
>+} AVD3D12VASyncContext;
>+
>+/**
>+ * @brief D3D12VA frame descriptor for pool allocation.
>+ *
>+ */
>+typedef struct AVD3D12VAFrame {
>+    /**
>+     * The texture in which the frame is located. The reference count is
>+     * managed by the AVBufferRef, and destroying the reference will release
>+     * the interface.
>+     */
>+    ID3D12Resource *texture;
>+
>+    /**
>+     * The index into the array texture element representing the frame
>+     */
>+    intptr_t index;
>+
>+    /**
>+     * The sync context for the texture
>+     *
>+     * Use av_d3d12va_wait_idle(sync_ctx) to ensure the decoding or encoding
>have been finised
>+     * @see: https://learn.microsoft.com/en-
>us/windows/win32/medfound/direct3d-12-video-overview#directx-12-fences
>+     */
>+    AVD3D12VASyncContext *sync_ctx;
>+} AVD3D12VAFrame;
>+
>+/**
>+ * @brief This struct is allocated as AVHWFramesContext.hwctx
>+ *
>+ */
>+typedef struct AVD3D12VAFramesContext {
>+    /**
>+     * This field is not able to be user-allocated at the present.
>+     */
>+    AVD3D12VAFrame *texture_infos;
>+} AVD3D12VAFramesContext;
>+
>+/**
>+ * @brief Map sw pixel format to d3d12 format
>+ *
>+ * @return d3d12 specified format
>+ */
>+DXGI_FORMAT av_d3d12va_map_sw_to_hw_format(enum AVPixelFormat
>pix_fmt);
>+
>+/**
>+ * @brief Allocate an AVD3D12VASyncContext
>+ *
>+ * @return Error code (ret < 0 if failed)
>+ */
>+int av_d3d12va_sync_context_alloc(AVD3D12VADeviceContext *ctx,
>AVD3D12VASyncContext **sync_ctx);
>+
>+/**
>+ * @brief Free an AVD3D12VASyncContext
>+ */
>+void av_d3d12va_sync_context_free(AVD3D12VASyncContext **sync_ctx);
>+
>+#endif /* AVUTIL_HWCONTEXT_D3D12VA_H */
>diff --git a/libavutil/hwcontext_d3d12va_internal.h
>b/libavutil/hwcontext_d3d12va_internal.h
>new file mode 100644
>index 0000000000..bfd89b3545
>--- /dev/null
>+++ b/libavutil/hwcontext_d3d12va_internal.h
>@@ -0,0 +1,59 @@
>+/*
>+ * Direct3D 12 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
>+ */
>+
>+#ifndef AVUTIL_HWCONTEXT_D3D12VA_INTERNAL_H
>+#define AVUTIL_HWCONTEXT_D3D12VA_INTERNAL_H
>+
>+/**
>+ * @def COBJMACROS
>+ *
>+ * @brief Enable C style interface for D3D12
>+ */
>+#ifndef COBJMACROS
>+#define COBJMACROS
>+#endif
>+
>+/**
>+ * @def DX_CHECK
>+ *
>+ * @brief A check macro used by D3D12 functions highly frequently
>+ */
>+#define DX_CHECK(hr)                              \
>+    do {                                          \
>+        if (FAILED(hr))                           \
>+            goto fail;                            \
>+    } while (0)
>+
>+/**
>+ * @def D3D12_OBJECT_RELEASE
>+ *
>+ * @brief A release macro used by D3D12 objects highly frequently
>+ */
>+#define D3D12_OBJECT_RELEASE(pInterface)              \
>+    do {                                              \
>+        if (pInterface) {                             \
>+            IUnknown_Release((IUnknown *)pInterface); \
>+            pInterface = NULL;                        \
>+        }                                             \
>+    } while (0)
>+
>+#endif /* AVUTIL_HWCONTEXT_D3D12VA_INTERNAL_H */
>\ No newline at end of file
>diff --git a/libavutil/hwcontext_internal.h b/libavutil/hwcontext_internal.h
>index e6266494ac..4df516ee6a 100644
>--- a/libavutil/hwcontext_internal.h
>+++ b/libavutil/hwcontext_internal.h
>@@ -165,6 +165,7 @@ int ff_hwframe_map_replace(AVFrame *dst, const
>AVFrame *src);
>
> extern const HWContextType ff_hwcontext_type_cuda;
> extern const HWContextType ff_hwcontext_type_d3d11va;
>+extern const HWContextType ff_hwcontext_type_d3d12va;
> extern const HWContextType ff_hwcontext_type_drm;
> extern const HWContextType ff_hwcontext_type_dxva2;
> extern const HWContextType ff_hwcontext_type_opencl;
>diff --git a/libavutil/pixdesc.c b/libavutil/pixdesc.c
>index e1e0dd2a9e..09fca5ef19 100644
>--- a/libavutil/pixdesc.c
>+++ b/libavutil/pixdesc.c
>@@ -2283,6 +2283,10 @@ static const AVPixFmtDescriptor
>av_pix_fmt_descriptors[AV_PIX_FMT_NB] = {
>         .name = "d3d11",
>         .flags = AV_PIX_FMT_FLAG_HWACCEL,
>     },
>+    [AV_PIX_FMT_D3D12] = {
>+        .name = "d3d12",
>+        .flags = AV_PIX_FMT_FLAG_HWACCEL,
>+    },
>     [AV_PIX_FMT_GBRPF32BE] = {
>         .name = "gbrpf32be",
>         .nb_components = 3,
>diff --git a/libavutil/pixfmt.h b/libavutil/pixfmt.h
>index 63e07ba64f..edf31cb574 100644
>--- a/libavutil/pixfmt.h
>+++ b/libavutil/pixfmt.h
>@@ -426,6 +426,13 @@ enum AVPixelFormat {
>     AV_PIX_FMT_P412BE,      ///< interleaved chroma YUV 4:4:4, 36bpp, data in
>the high bits, big-endian
>     AV_PIX_FMT_P412LE,      ///< interleaved chroma YUV 4:4:4, 36bpp, data in
>the high bits, little-endian
>
>+    /**
>+     * Hardware surfaces for Direct3D 12.
>+     *
>+     * data[0] points to an AVD3D12VAFrame
>+     */
>+    AV_PIX_FMT_D3D12,
>+
>     AV_PIX_FMT_NB         ///< number of pixel formats, DO NOT USE THIS if you
>want to link with shared libav* because the number of formats might differ
>between versions
> };
>
>diff --git a/libavutil/tests/hwdevice.c b/libavutil/tests/hwdevice.c
>index c57586613a..9d7964f9ee 100644
>--- a/libavutil/tests/hwdevice.c
>+++ b/libavutil/tests/hwdevice.c
>@@ -137,6 +137,8 @@ static const struct {
>       { "0", "1", "2" } },
>     { AV_HWDEVICE_TYPE_D3D11VA,
>       { "0", "1", "2" } },
>+    { AV_HWDEVICE_TYPE_D3D12VA,
>+      { "0", "1", "2" } },
>     { AV_HWDEVICE_TYPE_OPENCL,
>       { "0.0", "0.1", "1.0", "1.1" } },
>     { AV_HWDEVICE_TYPE_VAAPI,
>diff --git a/libavutil/version.h b/libavutil/version.h
>index bc43baca9f..6d27f91cce 100644
>--- a/libavutil/version.h
>+++ b/libavutil/version.h
>@@ -79,7 +79,7 @@
>  */
>
> #define LIBAVUTIL_VERSION_MAJOR  58
>-#define LIBAVUTIL_VERSION_MINOR  17
>+#define LIBAVUTIL_VERSION_MINOR  18
> #define LIBAVUTIL_VERSION_MICRO 100
>
> #define LIBAVUTIL_VERSION_INT
>AV_VERSION_INT(LIBAVUTIL_VERSION_MAJOR, \
>--
>2.41.0.windows.1

Aug 29, 2023, 06:15 by tong1.wu-at-intel.com@ffmpeg.org:

> +
> +    /**
> +     * Specifed by sync=1 when init d3d12va
> +     *
> +     * Execute commands as sync mode
> +     */
> +    int sync;
>

This is not needed, particularly in the public API.

> +    if (download) {
> +        ID3D12GraphicsCommandList_ResourceBarrier(s->command_list, 1, &barrier);
> +
> +        ID3D12GraphicsCommandList_CopyTextureRegion(s->command_list,
> +            &staging_y_location, 0, 0, 0, &texture_y_location, NULL);
> +
> +        ID3D12GraphicsCommandList_CopyTextureRegion(s->command_list,
> +            &staging_uv_location, 0, 0, 0, &texture_uv_location, NULL);
> +
> +        barrier.Transition.StateBefore = barrier.Transition.StateAfter;
> +        barrier.Transition.StateAfter = D3D12_RESOURCE_STATE_COMMON;
> +        ID3D12GraphicsCommandList_ResourceBarrier(s->command_list, 1, &barrier);
> +
> +        DX_CHECK(ID3D12GraphicsCommandList_Close(s->command_list));
> +
> +        if (!hwctx->sync)
> +            DX_CHECK(ID3D12CommandQueue_Wait(s->command_queue, sync_ctx->fence, sync_ctx->fence_value));

This is wrong. When downloading to RAM, the frames must
always be available and fully downloaded by the time the function
returns. Therefore, a wait must always occur.
Since this is the only place where sync is used, better remove the
option altogether.

>Aug 29, 2023, 06:15 by tong1.wu-at-intel.com@ffmpeg.org:
>
>> +
>> +    /**
>> +     * Specifed by sync=1 when init d3d12va
>> +     *
>> +     * Execute commands as sync mode
>> +     */
>> +    int sync;
>>
>
>This is not needed, particularly in the public API.
>
>> +    if (download) {
>> +        ID3D12GraphicsCommandList_ResourceBarrier(s->command_list, 1,
>&barrier);
>> +
>> +        ID3D12GraphicsCommandList_CopyTextureRegion(s->command_list,
>> +            &staging_y_location, 0, 0, 0, &texture_y_location, NULL);
>> +
>> +        ID3D12GraphicsCommandList_CopyTextureRegion(s->command_list,
>> +            &staging_uv_location, 0, 0, 0, &texture_uv_location, NULL);
>> +
>> +        barrier.Transition.StateBefore = barrier.Transition.StateAfter;
>> +        barrier.Transition.StateAfter = D3D12_RESOURCE_STATE_COMMON;
>> +        ID3D12GraphicsCommandList_ResourceBarrier(s->command_list, 1,
>&barrier);
>> +
>> +        DX_CHECK(ID3D12GraphicsCommandList_Close(s->command_list));
>> +
>> +        if (!hwctx->sync)
>> +            DX_CHECK(ID3D12CommandQueue_Wait(s->command_queue,
>sync_ctx->fence, sync_ctx->fence_value));
>
>This is wrong. When downloading to RAM, the frames must
>always be available and fully downloaded by the time the function
>returns. Therefore, a wait must always occur.
>Since this is the only place where sync is used, better remove the
>option altogether.

There's another place where it's used. It's designed for API users to blocking wait every decoded frame.
In end_frame() call in d3d12va_decode.c. 

    if (ctx->device_ctx->sync) {
        ret = d3d12va_wait_idle(ctx->sync_ctx);
        if (ret < 0)
            return ret;
    }

That's why when downloading, you don't have to wait again if hwctx->sync is specified. The frame has already been blocking synced in decoding process.

But yes I agree it's not that needed. I'll remove the whole design in next version.

Thanks,
Tong


>_______________________________________________
>ffmpeg-devel mailing list
>ffmpeg-devel@ffmpeg.org
>https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
>
>To unsubscribe, visit link above, or email
>ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".

Aug 30, 2023, 04:27 by tong1.wu-at-intel.com@ffmpeg.org:

> >Aug 29, 2023, 06:15 by tong1.wu-at-intel.com@ffmpeg.org:
>
>>> +
>>> +    /**
>>> +     * Specifed by sync=1 when init d3d12va
>>> +     *
>>> +     * Execute commands as sync mode
>>> +     */
>>> +    int sync;
>>>
> >This is not needed, particularly in the public API.
>
>>> +    if (download) {
>>> +        ID3D12GraphicsCommandList_ResourceBarrier(s->command_list, 1,
>>>
> >&barrier);
>
>>> +
>>> +        ID3D12GraphicsCommandList_CopyTextureRegion(s->command_list,
>>> +            &staging_y_location, 0, 0, 0, &texture_y_location, NULL);
>>> +
>>> +        ID3D12GraphicsCommandList_CopyTextureRegion(s->command_list,
>>> +            &staging_uv_location, 0, 0, 0, &texture_uv_location, NULL);
>>> +
>>> +        barrier.Transition.StateBefore = barrier.Transition.StateAfter;
>>> +        barrier.Transition.StateAfter = D3D12_RESOURCE_STATE_COMMON;
>>> +        ID3D12GraphicsCommandList_ResourceBarrier(s->command_list, 1,
>>>
> >&barrier);
>
>>> +
>>> +        DX_CHECK(ID3D12GraphicsCommandList_Close(s->command_list));
>>> +
>>> +        if (!hwctx->sync)
>>> +            DX_CHECK(ID3D12CommandQueue_Wait(s->command_queue,
>>>
> >sync_ctx->fence, sync_ctx->fence_value));
>
>>
>>
> >This is wrong. When downloading to RAM, the frames must
> >always be available and fully downloaded by the time the function
> >returns. Therefore, a wait must always occur.
> >Since this is the only place where sync is used, better remove the
> >option altogether.
>
> There's another place where it's used. It's designed for API users to blocking wait every decoded frame.
> In end_frame() call in d3d12va_decode.c. 
>
>  if (ctx->device_ctx->sync) {
>  ret = d3d12va_wait_idle(ctx->sync_ctx);
>  if (ret < 0)
>  return ret;
>  }
>
> That's why when downloading, you don't have to wait again if hwctx->sync is specified. The frame has already been blocking synced in decoding process.
>
> But yes I agree it's not that needed. I'll remove the whole design in next version.
>

That's not correct either, waiting should happen at the last possible moment,
rather than immediately after decoding, and waiting should happen asynchronously
on the GPU, rather than the CPU.

Would you mind holding off on posting the redesign until v6.1 is tagged?
Thanks.

>-----Original Message-----
>From: ffmpeg-devel <ffmpeg-devel-bounces@ffmpeg.org> On Behalf Of Lynne
>Sent: Wednesday, August 30, 2023 4:39 PM
>To: FFmpeg development discussions and patches <ffmpeg-
>devel@ffmpeg.org>
>Subject: Re: [FFmpeg-devel] [PATCH v7 1/9] libavutil: add hwcontext_d3d12va
>and AV_PIX_FMT_D3D12
>
>Aug 30, 2023, 04:27 by tong1.wu-at-intel.com@ffmpeg.org:
>
>> >Aug 29, 2023, 06:15 by tong1.wu-at-intel.com@ffmpeg.org:
>>
>>>> +
>>>> +    /**
>>>> +     * Specifed by sync=1 when init d3d12va
>>>> +     *
>>>> +     * Execute commands as sync mode
>>>> +     */
>>>> +    int sync;
>>>>
>> >This is not needed, particularly in the public API.
>>
>>>> +    if (download) {
>>>> +        ID3D12GraphicsCommandList_ResourceBarrier(s->command_list, 1,
>>>>
>> >&barrier);
>>
>>>> +
>>>> +        ID3D12GraphicsCommandList_CopyTextureRegion(s->command_list,
>>>> +            &staging_y_location, 0, 0, 0, &texture_y_location, NULL);
>>>> +
>>>> +        ID3D12GraphicsCommandList_CopyTextureRegion(s->command_list,
>>>> +            &staging_uv_location, 0, 0, 0, &texture_uv_location, NULL);
>>>> +
>>>> +        barrier.Transition.StateBefore = barrier.Transition.StateAfter;
>>>> +        barrier.Transition.StateAfter = D3D12_RESOURCE_STATE_COMMON;
>>>> +        ID3D12GraphicsCommandList_ResourceBarrier(s->command_list, 1,
>>>>
>> >&barrier);
>>
>>>> +
>>>> +        DX_CHECK(ID3D12GraphicsCommandList_Close(s->command_list));
>>>> +
>>>> +        if (!hwctx->sync)
>>>> +            DX_CHECK(ID3D12CommandQueue_Wait(s->command_queue,
>>>>
>> >sync_ctx->fence, sync_ctx->fence_value));
>>
>>>
>>>
>> >This is wrong. When downloading to RAM, the frames must
>> >always be available and fully downloaded by the time the function
>> >returns. Therefore, a wait must always occur.
>> >Since this is the only place where sync is used, better remove the
>> >option altogether.
>>
>> There's another place where it's used. It's designed for API users to blocking
>wait every decoded frame.
>> In end_frame() call in d3d12va_decode.c.
>>
>>  if (ctx->device_ctx->sync) {
>>  ret = d3d12va_wait_idle(ctx->sync_ctx);
>>  if (ret < 0)
>>  return ret;
>>  }
>>
>> That's why when downloading, you don't have to wait again if hwctx->sync is
>specified. The frame has already been blocking synced in decoding process.
>>
>> But yes I agree it's not that needed. I'll remove the whole design in next
>version.
>>
>
>That's not correct either, waiting should happen at the last possible moment,
>rather than immediately after decoding, and waiting should happen
>asynchronously
>on the GPU, rather than the CPU.
>
>Would you mind holding off on posting the redesign until v6.1 is tagged?
>Thanks.

I've already sent the v8 with the removal of the design. But yes it's ok for me to wait for 6.1 tagged and rebase again after that. BTW is there any schedule for 6.1? Haven't seen a lot of discussions on that recently. Thanks.

>_______________________________________________
>ffmpeg-devel mailing list
>ffmpeg-devel@ffmpeg.org
>https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
>
>To unsubscribe, visit link above, or email
>ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".

Aug 30, 2023, 11:05 by tong1.wu-at-intel.com@ffmpeg.org:

>
>
> >-----Original Message-----
> >From: ffmpeg-devel <ffmpeg-devel-bounces@ffmpeg.org> On Behalf Of Lynne
> >Sent: Wednesday, August 30, 2023 4:39 PM
> >To: FFmpeg development discussions and patches <ffmpeg-
> >devel@ffmpeg.org>
> >Subject: Re: [FFmpeg-devel] [PATCH v7 1/9] libavutil: add hwcontext_d3d12va
> >and AV_PIX_FMT_D3D12
>
>>
>>
> >Aug 30, 2023, 04:27 by tong1.wu-at-intel.com@ffmpeg.org:
>
>>> >Aug 29, 2023, 06:15 by tong1.wu-at-intel.com@ffmpeg.org:
>>>
>>>>> +
>>>>> +    /**
>>>>> +     * Specifed by sync=1 when init d3d12va
>>>>> +     *
>>>>> +     * Execute commands as sync mode
>>>>> +     */
>>>>> +    int sync;
>>>>>
>>> >This is not needed, particularly in the public API.
>>>
>>>>> +    if (download) {
>>>>> +        ID3D12GraphicsCommandList_ResourceBarrier(s->command_list, 1,
>>>>>
>>> >&barrier);
>>>
>>>>> +
>>>>> +        ID3D12GraphicsCommandList_CopyTextureRegion(s->command_list,
>>>>> +            &staging_y_location, 0, 0, 0, &texture_y_location, NULL);
>>>>> +
>>>>> +        ID3D12GraphicsCommandList_CopyTextureRegion(s->command_list,
>>>>> +            &staging_uv_location, 0, 0, 0, &texture_uv_location, NULL);
>>>>> +
>>>>> +        barrier.Transition.StateBefore = barrier.Transition.StateAfter;
>>>>> +        barrier.Transition.StateAfter = D3D12_RESOURCE_STATE_COMMON;
>>>>> +        ID3D12GraphicsCommandList_ResourceBarrier(s->command_list, 1,
>>>>>
>>> >&barrier);
>>>
>>>>> +
>>>>> +        DX_CHECK(ID3D12GraphicsCommandList_Close(s->command_list));
>>>>> +
>>>>> +        if (!hwctx->sync)
>>>>> +            DX_CHECK(ID3D12CommandQueue_Wait(s->command_queue,
>>>>>
>>> >sync_ctx->fence, sync_ctx->fence_value));
>>>
>>>>
>>>>
>>> >This is wrong. When downloading to RAM, the frames must
>>> >always be available and fully downloaded by the time the function
>>> >returns. Therefore, a wait must always occur.
>>> >Since this is the only place where sync is used, better remove the
>>> >option altogether.
>>>
>>> There's another place where it's used. It's designed for API users to blocking
>>>
> >wait every decoded frame.
>
>>> In end_frame() call in d3d12va_decode.c.
>>>
>>>  if (ctx->device_ctx->sync) {
>>>  ret = d3d12va_wait_idle(ctx->sync_ctx);
>>>  if (ret < 0)
>>>  return ret;
>>>  }
>>>
>>> That's why when downloading, you don't have to wait again if hwctx->sync is
>>>
> >specified. The frame has already been blocking synced in decoding process.
>
>>>
>>> But yes I agree it's not that needed. I'll remove the whole design in next
>>>
> >version.
>
>>>
>>>
> >That's not correct either, waiting should happen at the last possible moment,
> >rather than immediately after decoding, and waiting should happen
> >asynchronously
> >on the GPU, rather than the CPU.
>
>>
>>
> >Would you mind holding off on posting the redesign until v6.1 is tagged?
> >Thanks.
>
> I've already sent the v8 with the removal of the design.
>

The code is still present in v8 though, the d3d12va_wait_idle() call
in ff_d3d12va_common_end_frame. Is there no way to wait for
the operation on the GPU by recording it in the command buffer?


>  But yes it's ok for me to wait for 6.1 tagged and rebase again after that. BTW is there any schedule for 6.1? Haven't seen a lot of discussions on that recently. Thanks.
>

Thanks.

>Aug 30, 2023, 11:05 by tong1.wu-at-intel.com@ffmpeg.org:
>
>>
>>
>> >-----Original Message-----
>> >From: ffmpeg-devel <ffmpeg-devel-bounces@ffmpeg.org> On Behalf Of
>Lynne
>> >Sent: Wednesday, August 30, 2023 4:39 PM
>> >To: FFmpeg development discussions and patches <ffmpeg-
>> >devel@ffmpeg.org>
>> >Subject: Re: [FFmpeg-devel] [PATCH v7 1/9] libavutil: add
>hwcontext_d3d12va
>> >and AV_PIX_FMT_D3D12
>>
>>>
>>>
>> >Aug 30, 2023, 04:27 by tong1.wu-at-intel.com@ffmpeg.org:
>>
>>>> >Aug 29, 2023, 06:15 by tong1.wu-at-intel.com@ffmpeg.org:
>>>>
>>>>>> +
>>>>>> +    /**
>>>>>> +     * Specifed by sync=1 when init d3d12va
>>>>>> +     *
>>>>>> +     * Execute commands as sync mode
>>>>>> +     */
>>>>>> +    int sync;
>>>>>>
>>>> >This is not needed, particularly in the public API.
>>>>
>>>>>> +    if (download) {
>>>>>> +        ID3D12GraphicsCommandList_ResourceBarrier(s->command_list,
>1,
>>>>>>
>>>> >&barrier);
>>>>
>>>>>> +
>>>>>> +        ID3D12GraphicsCommandList_CopyTextureRegion(s-
>>command_list,
>>>>>> +            &staging_y_location, 0, 0, 0, &texture_y_location, NULL);
>>>>>> +
>>>>>> +        ID3D12GraphicsCommandList_CopyTextureRegion(s-
>>command_list,
>>>>>> +            &staging_uv_location, 0, 0, 0, &texture_uv_location, NULL);
>>>>>> +
>>>>>> +        barrier.Transition.StateBefore = barrier.Transition.StateAfter;
>>>>>> +        barrier.Transition.StateAfter =
>D3D12_RESOURCE_STATE_COMMON;
>>>>>> +        ID3D12GraphicsCommandList_ResourceBarrier(s->command_list,
>1,
>>>>>>
>>>> >&barrier);
>>>>
>>>>>> +
>>>>>> +        DX_CHECK(ID3D12GraphicsCommandList_Close(s-
>>command_list));
>>>>>> +
>>>>>> +        if (!hwctx->sync)
>>>>>> +            DX_CHECK(ID3D12CommandQueue_Wait(s->command_queue,
>>>>>>
>>>> >sync_ctx->fence, sync_ctx->fence_value));
>>>>
>>>>>
>>>>>
>>>> >This is wrong. When downloading to RAM, the frames must
>>>> >always be available and fully downloaded by the time the function
>>>> >returns. Therefore, a wait must always occur.
>>>> >Since this is the only place where sync is used, better remove the
>>>> >option altogether.
>>>>
>>>> There's another place where it's used. It's designed for API users to
>blocking
>>>>
>> >wait every decoded frame.
>>
>>>> In end_frame() call in d3d12va_decode.c.
>>>>
>>>>  if (ctx->device_ctx->sync) {
>>>>  ret = d3d12va_wait_idle(ctx->sync_ctx);
>>>>  if (ret < 0)
>>>>  return ret;
>>>>  }
>>>>
>>>> That's why when downloading, you don't have to wait again if hwctx->sync
>is
>>>>
>> >specified. The frame has already been blocking synced in decoding process.
>>
>>>>
>>>> But yes I agree it's not that needed. I'll remove the whole design in next
>>>>
>> >version.
>>
>>>>
>>>>
>> >That's not correct either, waiting should happen at the last possible
>moment,
>> >rather than immediately after decoding, and waiting should happen
>> >asynchronously
>> >on the GPU, rather than the CPU.
>>
>>>
>>>
>> >Would you mind holding off on posting the redesign until v6.1 is tagged?
>> >Thanks.
>>
>> I've already sent the v8 with the removal of the design.
>>
>
>The code is still present in v8 though, the d3d12va_wait_idle() call
>in ff_d3d12va_common_end_frame. Is there no way to wait for
>the operation on the GPU by recording it in the command buffer?

Actually d3d12va_wait_idle is just a query of timeline fence. At most of time, the block wait is not triggered. It only gets the fence value to know if the GPU command is completed. It's a known d3d12 design for high performance and for users to sync with GPU completion. 

The function name might be a bit confusing. I may change the name from "wait" to "sync" in the next version.

Thanks,
Tong

>
>
>>  But yes it's ok for me to wait for 6.1 tagged and rebase again after that. BTW
>is there any schedule for 6.1? Haven't seen a lot of discussions on that recently.
>Thanks.
>>
>
>Thanks.
>_______________________________________________
>ffmpeg-devel mailing list
>ffmpeg-devel@ffmpeg.org
>https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
>
>To unsubscribe, visit link above, or email
>ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".