Before 741b494fa8cd28a7d096349bac183893c236e3f9, when the reference list
modification description was invalid, the code would substitute the
corresponding reference from the initial ("default") reference list.
After that commit, it will just return an error.Since there are apparently invalid samples in the wild that used to play fine with the old code, it is a good idea to re-add some sort of error resilience here. So, when the reference list modification results in a missing frame, substitute a previous reference frame for it. The relevant sample again decodes fine with the same output as previously. --- Thanks to Martin for telling me about the problem. The sample is http://files.1f0.de/samples/h264refframeregression.mp4 I wonder if we should add a FATE test for this. The file is invalid, but we probably still want it to remaing working. --- libavcodec/h264_refs.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/libavcodec/h264_refs.c b/libavcodec/h264_refs.c index 46b51e0..877f6fe 100644 --- a/libavcodec/h264_refs.c +++ b/libavcodec/h264_refs.c @@ -325,7 +325,10 @@ int ff_h264_decode_ref_pic_list_reordering(H264Context *h, H264SliceContext *sl) for (index = 0; index < sl->ref_count[list]; index++) { if (!sl->ref_list[list][index].parent) { av_log(h->avctx, AV_LOG_ERROR, "Missing reference picture\n"); - return -1; + if (index == 0 || h->avctx->err_recognition & AV_EF_EXPLODE) + return AVERROR_INVALIDDATA; + else + sl->ref_list[list][index] = sl->ref_list[list][index - 1]; } } } -- 2.0.0 _______________________________________________ libav-devel mailing list [email protected] https://lists.libav.org/mailman/listinfo/libav-devel
