int ff_reget_buffer(AVCodecContext *avctx, AVFrame *frame)
{
AVFrame *tmp;
int ret;
av_assert0(avctx->codec_type == AVMEDIA_TYPE_VIDEO);
if (!frame->data[0])
return ff_get_buffer(avctx, frame, AV_GET_BUFFER_FLAG_REF);
if (av_frame_is_writable(frame))
return ff_decode_frame_props(avctx, frame);
tmp = av_frame_alloc();
if (!tmp)
return AVERROR(ENOMEM);
av_frame_move_ref(tmp, frame);
ret = ff_get_buffer(avctx, frame, AV_GET_BUFFER_FLAG_REF);
if (ret < 0) {
av_frame_free(&tmp);
return ret;
}
av_frame_copy(frame, tmp);
av_frame_free(&tmp);
return 0;
}
static int ffmal_copy_frame(AVCodecContext *avctx, AVFrame *frame,
MMAL_BUFFER_HEADER_T *buffer)
{
MMALDecodeContext *ctx = avctx->priv_data;
int ret = 0;
if (avctx->pix_fmt == AV_PIX_FMT_MMAL) {
if (!ctx->pool_out)
return AVERROR_UNKNOWN; // format change code failed with OOM previously
if ((ret = ff_decode_frame_props(avctx, frame)) < 0)
goto done;
if ((ret = ffmmal_set_ref(frame, ctx->pool_out, buffer)) < 0)
goto done;
} else {
int w = FFALIGN(avctx->width, 32);
int h = FFALIGN(avctx->height, 16);
char *ptr;
int plane;
int i;
if ((ret = ff_get_buffer(avctx, frame, 0)) < 0)
goto done;
ptr = buffer->data + buffer->type->video.offset[0];
for (i = 0; i < avctx->height; i++)
memcpy(frame->data[0] + frame->linesize[0] * i, ptr + w * i, avctx->width);
ptr += w * h;
for (plane = 1; plane < 3; plane++) {
for (i = 0; i < avctx->height / 2; i++)
memcpy(frame->data[plane] + frame->linesize[plane] * i, ptr + w / 2 * i, (avctx->width + 1) / 2);
ptr += w / 2 * h / 2;
}
}
if (buffer->pts != MMAL_TIME_UNKNOWN) {
frame->pkt_pts = buffer->pts;
frame->pts = buffer->pts;
}
done:
return ret;
}
static int ffmal_copy_frame(AVCodecContext *avctx, AVFrame *frame,
MMAL_BUFFER_HEADER_T *buffer)
{
MMALDecodeContext *ctx = avctx->priv_data;
int ret = 0;
if (avctx->pix_fmt == AV_PIX_FMT_MMAL) {
if (!ctx->pool_out)
return AVERROR_UNKNOWN; // format change code failed with OOM previously
if ((ret = ff_decode_frame_props(avctx, frame)) < 0)
goto done;
if ((ret = ffmmal_set_ref(frame, ctx->pool_out, buffer)) < 0)
goto done;
} else {
int w = FFALIGN(avctx->width, 32);
int h = FFALIGN(avctx->height, 16);
uint8_t *src[4];
int linesize[4];
if ((ret = ff_get_buffer(avctx, frame, 0)) < 0)
goto done;
av_image_fill_arrays(src, linesize,
buffer->data + buffer->type->video.offset[0],
avctx->pix_fmt, w, h, 1);
av_image_copy(frame->data, frame->linesize, src, linesize,
avctx->pix_fmt, avctx->width, avctx->height);
}
frame->pts = buffer->pts == MMAL_TIME_UNKNOWN ? AV_NOPTS_VALUE : buffer->pts;
#if FF_API_PKT_PTS
FF_DISABLE_DEPRECATION_WARNINGS
frame->pkt_pts = frame->pts;
FF_ENABLE_DEPRECATION_WARNINGS
#endif
frame->pkt_dts = AV_NOPTS_VALUE;
done:
return ret;
}
static int cuvid_output_frame(AVCodecContext *avctx, AVFrame *frame)
{
CuvidContext *ctx = avctx->priv_data;
AVHWDeviceContext *device_ctx = (AVHWDeviceContext*)ctx->hwdevice->data;
AVCUDADeviceContext *device_hwctx = device_ctx->hwctx;
CUcontext dummy, cuda_ctx = device_hwctx->cuda_ctx;
CUdeviceptr mapped_frame = 0;
int ret = 0, eret = 0;
av_log(avctx, AV_LOG_TRACE, "cuvid_output_frame\n");
if (ctx->decoder_flushing) {
ret = cuvid_decode_packet(avctx, NULL);
if (ret < 0 && ret != AVERROR_EOF)
return ret;
}
ret = CHECK_CU(cuCtxPushCurrent(cuda_ctx));
if (ret < 0)
return ret;
if (av_fifo_size(ctx->frame_queue)) {
CuvidParsedFrame parsed_frame;
CUVIDPROCPARAMS params;
unsigned int pitch = 0;
int offset = 0;
int i;
av_fifo_generic_read(ctx->frame_queue, &parsed_frame, sizeof(CuvidParsedFrame), NULL);
memset(¶ms, 0, sizeof(params));
params.progressive_frame = parsed_frame.dispinfo.progressive_frame;
params.second_field = parsed_frame.second_field;
params.top_field_first = parsed_frame.dispinfo.top_field_first;
ret = CHECK_CU(cuvidMapVideoFrame(ctx->cudecoder, parsed_frame.dispinfo.picture_index, &mapped_frame, &pitch, ¶ms));
if (ret < 0)
goto error;
if (avctx->pix_fmt == AV_PIX_FMT_CUDA) {
ret = av_hwframe_get_buffer(ctx->hwframe, frame, 0);
if (ret < 0) {
av_log(avctx, AV_LOG_ERROR, "av_hwframe_get_buffer failed\n");
goto error;
}
ret = ff_decode_frame_props(avctx, frame);
if (ret < 0) {
av_log(avctx, AV_LOG_ERROR, "ff_decode_frame_props failed\n");
goto error;
}
for (i = 0; i < 2; i++) {
CUDA_MEMCPY2D cpy = {
.srcMemoryType = CU_MEMORYTYPE_DEVICE,
.dstMemoryType = CU_MEMORYTYPE_DEVICE,
.srcDevice = mapped_frame,
.dstDevice = (CUdeviceptr)frame->data[i],
.srcPitch = pitch,
.dstPitch = frame->linesize[i],
.srcY = offset,
.WidthInBytes = FFMIN(pitch, frame->linesize[i]),
.Height = avctx->height >> (i ? 1 : 0),
};
ret = CHECK_CU(cuMemcpy2D(&cpy));
if (ret < 0)
goto error;
offset += avctx->coded_height;
}
} else if (avctx->pix_fmt == AV_PIX_FMT_NV12) {
int ff_get_buffer(AVCodecContext *avctx, AVFrame *frame, int flags)
{
const AVHWAccel *hwaccel = avctx->hwaccel;
int override_dimensions = 1;
int ret;
switch (avctx->codec_type) {
case AVMEDIA_TYPE_VIDEO:
if (frame->width <= 0 || frame->height <= 0) {
frame->width = FFMAX(avctx->width, avctx->coded_width);
frame->height = FFMAX(avctx->height, avctx->coded_height);
override_dimensions = 0;
}
if (frame->format < 0)
frame->format = avctx->pix_fmt;
if (!frame->sample_aspect_ratio.num)
frame->sample_aspect_ratio = avctx->sample_aspect_ratio;
if (av_image_check_sar(frame->width, frame->height,
frame->sample_aspect_ratio) < 0) {
av_log(avctx, AV_LOG_WARNING, "ignoring invalid SAR: %u/%u\n",
frame->sample_aspect_ratio.num,
frame->sample_aspect_ratio.den);
frame->sample_aspect_ratio = (AVRational){ 0, 1 };
}
if ((ret = av_image_check_size(avctx->width, avctx->height, 0, avctx)) < 0)
return ret;
break;
case AVMEDIA_TYPE_AUDIO:
if (!frame->sample_rate)
frame->sample_rate = avctx->sample_rate;
if (frame->format < 0)
frame->format = avctx->sample_fmt;
if (!frame->channel_layout) {
if (avctx->channel_layout) {
if (av_get_channel_layout_nb_channels(avctx->channel_layout) !=
avctx->channels) {
av_log(avctx, AV_LOG_ERROR, "Inconsistent channel "
"configuration.\n");
return AVERROR(EINVAL);
}
frame->channel_layout = avctx->channel_layout;
} else {
if (avctx->channels > FF_SANE_NB_CHANNELS) {
av_log(avctx, AV_LOG_ERROR, "Too many channels: %d.\n",
avctx->channels);
return AVERROR(ENOSYS);
}
frame->channel_layout = av_get_default_channel_layout(avctx->channels);
if (!frame->channel_layout)
frame->channel_layout = (1ULL << avctx->channels) - 1;
}
}
break;
default: return AVERROR(EINVAL);
}
ret = ff_decode_frame_props(avctx, frame);
if (ret < 0)
return ret;
if (hwaccel) {
if (hwaccel->alloc_frame) {
ret = hwaccel->alloc_frame(avctx, frame);
goto end;
}
} else
avctx->sw_pix_fmt = avctx->pix_fmt;
ret = avctx->get_buffer2(avctx, frame, flags);
if (ret < 0)
goto end;
ret = attach_decode_data(frame);
if (ret < 0)
goto end;
end:
if (avctx->codec_type == AVMEDIA_TYPE_VIDEO && !override_dimensions &&
!(avctx->codec->caps_internal & FF_CODEC_CAP_EXPORTS_CROPPING)) {
frame->width = avctx->width;
frame->height = avctx->height;
}
if (ret < 0)
av_frame_unref(frame);
return ret;
}
static int cuvid_output_frame(AVCodecContext *avctx, AVFrame *frame)
{
CuvidContext *ctx = avctx->priv_data;
AVHWDeviceContext *device_ctx = (AVHWDeviceContext*)ctx->hwdevice->data;
AVCUDADeviceContext *device_hwctx = device_ctx->hwctx;
CUcontext dummy, cuda_ctx = device_hwctx->cuda_ctx;
CUdeviceptr mapped_frame = 0;
int ret = 0, eret = 0;
av_log(avctx, AV_LOG_TRACE, "cuvid_output_frame\n");
if (ctx->decoder_flushing) {
ret = cuvid_decode_packet(avctx, NULL);
if (ret < 0 && ret != AVERROR_EOF)
return ret;
}
if (!cuvid_is_buffer_full(avctx)) {
AVPacket pkt = {0};
ret = ff_decode_get_packet(avctx, &pkt);
if (ret < 0 && ret != AVERROR_EOF)
return ret;
ret = cuvid_decode_packet(avctx, &pkt);
av_packet_unref(&pkt);
// cuvid_is_buffer_full() should avoid this.
if (ret == AVERROR(EAGAIN))
ret = AVERROR_EXTERNAL;
if (ret < 0 && ret != AVERROR_EOF)
return ret;
}
ret = CHECK_CU(ctx->cudl->cuCtxPushCurrent(cuda_ctx));
if (ret < 0)
return ret;
if (av_fifo_size(ctx->frame_queue)) {
const AVPixFmtDescriptor *pixdesc;
CuvidParsedFrame parsed_frame;
CUVIDPROCPARAMS params;
unsigned int pitch = 0;
int offset = 0;
int i;
av_fifo_generic_read(ctx->frame_queue, &parsed_frame, sizeof(CuvidParsedFrame), NULL);
memset(¶ms, 0, sizeof(params));
params.progressive_frame = parsed_frame.dispinfo.progressive_frame;
params.second_field = parsed_frame.second_field;
params.top_field_first = parsed_frame.dispinfo.top_field_first;
ret = CHECK_CU(ctx->cvdl->cuvidMapVideoFrame(ctx->cudecoder, parsed_frame.dispinfo.picture_index, &mapped_frame, &pitch, ¶ms));
if (ret < 0)
goto error;
if (avctx->pix_fmt == AV_PIX_FMT_CUDA) {
ret = av_hwframe_get_buffer(ctx->hwframe, frame, 0);
if (ret < 0) {
av_log(avctx, AV_LOG_ERROR, "av_hwframe_get_buffer failed\n");
goto error;
}
ret = ff_decode_frame_props(avctx, frame);
if (ret < 0) {
av_log(avctx, AV_LOG_ERROR, "ff_decode_frame_props failed\n");
goto error;
}
pixdesc = av_pix_fmt_desc_get(avctx->sw_pix_fmt);
for (i = 0; i < pixdesc->nb_components; i++) {
int height = avctx->height >> (i ? pixdesc->log2_chroma_h : 0);
CUDA_MEMCPY2D cpy = {
.srcMemoryType = CU_MEMORYTYPE_DEVICE,
.dstMemoryType = CU_MEMORYTYPE_DEVICE,
.srcDevice = mapped_frame,
.dstDevice = (CUdeviceptr)frame->data[i],
.srcPitch = pitch,
.dstPitch = frame->linesize[i],
.srcY = offset,
.WidthInBytes = FFMIN(pitch, frame->linesize[i]),
.Height = height,
};
ret = CHECK_CU(ctx->cudl->cuMemcpy2DAsync(&cpy, device_hwctx->stream));
if (ret < 0)
goto error;
offset += height;
}
} else if (avctx->pix_fmt == AV_PIX_FMT_NV12 ||
avctx->pix_fmt == AV_PIX_FMT_P010 ||
avctx->pix_fmt == AV_PIX_FMT_P016 ||
avctx->pix_fmt == AV_PIX_FMT_YUV444P ||
avctx->pix_fmt == AV_PIX_FMT_YUV444P16) {
unsigned int offset = 0;
AVFrame *tmp_frame = av_frame_alloc();
if (!tmp_frame) {
av_log(avctx, AV_LOG_ERROR, "av_frame_alloc failed\n");
ret = AVERROR(ENOMEM);
goto error;
}
pixdesc = av_pix_fmt_desc_get(avctx->sw_pix_fmt);
tmp_frame->format = AV_PIX_FMT_CUDA;
tmp_frame->hw_frames_ctx = av_buffer_ref(ctx->hwframe);
tmp_frame->width = avctx->width;
tmp_frame->height = avctx->height;
/*
* Note that the following logic would not work for three plane
* YUV420 because the pitch value is different for the chroma
* planes.
*/
for (i = 0; i < pixdesc->nb_components; i++) {
tmp_frame->data[i] = (uint8_t*)mapped_frame + offset;
tmp_frame->linesize[i] = pitch;
offset += pitch * (avctx->height >> (i ? pixdesc->log2_chroma_h : 0));
}
ret = ff_get_buffer(avctx, frame, 0);
if (ret < 0) {
av_log(avctx, AV_LOG_ERROR, "ff_get_buffer failed\n");
av_frame_free(&tmp_frame);
goto error;
}
ret = av_hwframe_transfer_data(frame, tmp_frame, 0);
if (ret) {
av_log(avctx, AV_LOG_ERROR, "av_hwframe_transfer_data failed\n");
av_frame_free(&tmp_frame);
goto error;
}
av_frame_free(&tmp_frame);
} else {
static int cuvid_decode_frame(AVCodecContext *avctx, void *data, int *got_frame, AVPacket *avpkt)
{
CuvidContext *ctx = avctx->priv_data;
AVHWDeviceContext *device_ctx = (AVHWDeviceContext*)ctx->hwdevice->data;
AVCUDADeviceContext *device_hwctx = device_ctx->hwctx;
CUcontext dummy, cuda_ctx = device_hwctx->cuda_ctx;
AVFrame *frame = data;
CUVIDSOURCEDATAPACKET cupkt;
AVPacket filter_packet = { 0 };
AVPacket filtered_packet = { 0 };
CUdeviceptr mapped_frame = 0;
int ret = 0, eret = 0;
if (ctx->bsf && avpkt->size) {
if ((ret = av_packet_ref(&filter_packet, avpkt)) < 0) {
av_log(avctx, AV_LOG_ERROR, "av_packet_ref failed\n");
return ret;
}
if ((ret = av_bsf_send_packet(ctx->bsf, &filter_packet)) < 0) {
av_log(avctx, AV_LOG_ERROR, "av_bsf_send_packet failed\n");
av_packet_unref(&filter_packet);
return ret;
}
if ((ret = av_bsf_receive_packet(ctx->bsf, &filtered_packet)) < 0) {
av_log(avctx, AV_LOG_ERROR, "av_bsf_receive_packet failed\n");
return ret;
}
avpkt = &filtered_packet;
}
ret = CHECK_CU(cuCtxPushCurrent(cuda_ctx));
if (ret < 0) {
av_packet_unref(&filtered_packet);
return ret;
}
memset(&cupkt, 0, sizeof(cupkt));
if (avpkt->size) {
cupkt.payload_size = avpkt->size;
cupkt.payload = avpkt->data;
if (avpkt->pts != AV_NOPTS_VALUE) {
cupkt.flags = CUVID_PKT_TIMESTAMP;
if (avctx->pkt_timebase.num && avctx->pkt_timebase.den)
cupkt.timestamp = av_rescale_q(avpkt->pts, avctx->pkt_timebase, (AVRational){1, 10000000});
else
cupkt.timestamp = avpkt->pts;
}
} else {
cupkt.flags = CUVID_PKT_ENDOFSTREAM;
}
ret = CHECK_CU(cuvidParseVideoData(ctx->cuparser, &cupkt));
av_packet_unref(&filtered_packet);
if (ret < 0) {
goto error;
}
// cuvidParseVideoData doesn't return an error just because stuff failed...
if (ctx->internal_error) {
av_log(avctx, AV_LOG_ERROR, "cuvid decode callback error\n");
ret = ctx->internal_error;
goto error;
}
if (av_fifo_size(ctx->frame_queue)) {
CUVIDPARSERDISPINFO dispinfo;
CUVIDPROCPARAMS params;
unsigned int pitch = 0;
int offset = 0;
int i;
av_fifo_generic_read(ctx->frame_queue, &dispinfo, sizeof(CUVIDPARSERDISPINFO), NULL);
memset(¶ms, 0, sizeof(params));
params.progressive_frame = dispinfo.progressive_frame;
params.second_field = 0;
params.top_field_first = dispinfo.top_field_first;
ret = CHECK_CU(cuvidMapVideoFrame(ctx->cudecoder, dispinfo.picture_index, &mapped_frame, &pitch, ¶ms));
if (ret < 0)
goto error;
if (avctx->pix_fmt == AV_PIX_FMT_CUDA) {
ret = av_hwframe_get_buffer(ctx->hwframe, frame, 0);
if (ret < 0) {
av_log(avctx, AV_LOG_ERROR, "av_hwframe_get_buffer failed\n");
goto error;
}
ret = ff_decode_frame_props(avctx, frame);
if (ret < 0) {
av_log(avctx, AV_LOG_ERROR, "ff_decode_frame_props failed\n");
goto error;
}
for (i = 0; i < 2; i++) {
CUDA_MEMCPY2D cpy = {
.srcMemoryType = CU_MEMORYTYPE_DEVICE,
.dstMemoryType = CU_MEMORYTYPE_DEVICE,
.srcDevice = mapped_frame,
.dstDevice = (CUdeviceptr)frame->data[i],
.srcPitch = pitch,
.dstPitch = frame->linesize[i],
.srcY = offset,
.WidthInBytes = FFMIN(pitch, frame->linesize[i]),
.Height = avctx->coded_height >> (i ? 1 : 0),
};
ret = CHECK_CU(cuMemcpy2D(&cpy));
if (ret < 0)
goto error;
offset += avctx->coded_height;
}
} else if (avctx->pix_fmt == AV_PIX_FMT_NV12) {