And I think that this memleak in mkv_write_trailer() has a twin in mkv_write_packet(): Audio frames are not written directly, but rather put into storage in cur_audio_pkt via av_packet_ref(). But if the preceding audio packet was a zero size packet (but possibly with side-data), then the preceding packet's side data will leak at this point. A better solution seems to be never to store a packet whose buffer has size zero in cur_audio_pkt. The only use we have for such packets is to use them for mkv_check_new_extra_data() and that has already been done at this point.
- Andreas _______________________________________________ ffmpeg-devel mailing list [email protected] https://ffmpeg.org/mailman/listinfo/ffmpeg-devel To unsubscribe, visit link above, or email [email protected] with subject "unsubscribe".
