Message ID | 20200522012431.29918-3-andreas.rheinhardt@gmail.com |
---|---|
State | Accepted |
Commit | 38490cbeb3a48dd488f10a72efc8d8e3f8883ac4 |
Headers | show |
Series | [FFmpeg-devel,1/3] avformat/matroskaenc: Don't use stream side-data size | expand |
Context | Check | Description |
---|---|---|
andriy/default | pending | |
andriy/make | success | Make finished |
andriy/make_fate | success | Make fate finished |
Quoting Andreas Rheinhardt (2020-05-22 03:24:31) > Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com> > --- > Was the earlier behaviour actually a bug (I think so) or is this new > behaviour something that should be explicitly documented? > > Btw, the stream and packet side-data APIs differ in two more ways: > av_packet_new_side_data() adds padding to the buffer and zeroes it > whereas av_stream_new_side_data() does not. Should this be unified? Given that the side data can be extradata, I'd say yes to padding. For zeroing, not sure. The caller should fill the data completely, so zeroing it may hide missed initialization.
diff --git a/libavformat/utils.c b/libavformat/utils.c index e6158d8058..b12aff5eb0 100644 --- a/libavformat/utils.c +++ b/libavformat/utils.c @@ -5507,6 +5507,8 @@ uint8_t *av_stream_get_side_data(const AVStream *st, return st->side_data[i].data; } } + if (size) + *size = 0; return NULL; }
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com> --- Was the earlier behaviour actually a bug (I think so) or is this new behaviour something that should be explicitly documented? Btw, the stream and packet side-data APIs differ in two more ways: av_packet_new_side_data() adds padding to the buffer and zeroes it whereas av_stream_new_side_data() does not. Should this be unified? libavformat/utils.c | 2 ++ 1 file changed, 2 insertions(+)