| Message ID | 20200530160541.29517-10-andreas.rheinhardt@gmail.com |
|---|---|
| State | New |
| Headers | show |
| Series | [FFmpeg-devel,01/36] avcodec/vp9_superframe_bsf: Check for existence of data before reading it | expand |
| Context | Check | Description |
|---|---|---|
| andriy/default | pending | |
| andriy/make | success | Make finished |
| andriy/make_fate | success | Make fate finished |
Quoting Andreas Rheinhardt (2020-05-30 18:05:15) > It has already been checked that there is enough data available. > > Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com> > --- I don't see how this improves things. The only advantage of unchecked reads is performance, but I don't think that is relevant here. Otherwise, it just adds a non-obvious dependency on the check being there and being correct (not to mention cluttering history). IMO unchecked reads should be used ONLY when there are important performance reasons to use them, otherwise checked reads should be used even if they are not strictly necessary.
diff --git a/libavcodec/vp9_superframe_split_bsf.c b/libavcodec/vp9_superframe_split_bsf.c index 48c3723206..9f9e495bd5 100644 --- a/libavcodec/vp9_superframe_split_bsf.c +++ b/libavcodec/vp9_superframe_split_bsf.c @@ -72,7 +72,7 @@ static int vp9_superframe_split_filter(AVBSFContext *ctx, AVPacket *out) for (i = 0; i < nb_frames; i++) { unsigned frame_size = 0; for (j = 0; j < length_size; j++) - frame_size |= bytestream2_get_byte(&bc) << (j * 8); + frame_size |= bytestream2_get_byteu(&bc) << (j * 8); total_size += frame_size; if (!frame_size || total_size > in->size - idx_size) {
It has already been checked that there is enough data available. Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com> --- libavcodec/vp9_superframe_split_bsf.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)