[FFmpeg-devel] [PATCH 2/3] lavc/qsvenc: avoid data copy if possible

[Xiang, Haihao]

From: Haihao Xiang <haihao.xiang at intel.com>

The data copy is unnecessary for packed formats when frame width and
height are aligned

For example:
$ ffmpeg -f lavfi -i testsrc=size=1920x1088 -vf "format=yuyv422" -c:v hevc_qsv -f null -

Signed-off-by: Haihao Xiang <haihao.xiang at intel.com>
---
 libavcodec/qsvenc.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/libavcodec/qsvenc.c b/libavcodec/qsvenc.c
index 2c38fbf0dc..b6813b3023 100644
--- a/libavcodec/qsvenc.c
+++ b/libavcodec/qsvenc.c
@@ -1910,7 +1910,8 @@ static int submit_frame(QSVEncContext *q, const AVFrame *frame,
         /* make a copy if the input is not padded as libmfx requires */
         /* and to make allocation continious for data[0]/data[1] */
          if ((frame->height & (q->height_align - 1) || frame->linesize[0] & (q->width_align - 1)) ||
-            (frame->data[1] - frame->data[0] != frame->linesize[0] * FFALIGN(qf->frame->height, q->height_align))) {
+            ((frame->format == AV_PIX_FMT_NV12 || frame->format == AV_PIX_FMT_P010 || frame->format == AV_PIX_FMT_P012) &&
+             (frame->data[1] - frame->data[0] != frame->linesize[0] * FFALIGN(qf->frame->height, q->height_align)))) {
             int tmp_w, tmp_h;
             qf->frame->height = tmp_h = FFALIGN(frame->height, q->height_align);
             qf->frame->width  = tmp_w = FFALIGN(frame->width, q->width_align);
-- 
2.34.1