#5994: Heap-overflow in rtmppkt.c results Remote Code Execution ----------------------------------+-------------------------------------- Reporter: paulch | Type: defect Status: new | Priority: critical Component: avformat | Version: git-master Keywords: | Blocked By: Blocking: | Reproduced by developer: 0 Analyzed by developer: 0 | ----------------------------------+-------------------------------------- '''Summary of the bug''': We managed to find and successfully exploit critical security bug in libavformat/rtmppkt.c that results Remote Code Execution in latest version of FFmpeg. '''Prerequisites:''' * Attacker has to reproduce SSRF bug and send RTMP request to his remote host. There are multiple ways of doing this already described all over the Internet. (For ex. you can trigger SSRF using HLS playlists or the way I described earlier in this ​[https://trac.ffmpeg.org/ticket/5991#ticket ticket] ). * For full exploitation knowledge of FFmpeg binary is required. '''Overview:''' Main issue is contained inside [https://github.com/FFmpeg/FFmpeg/blob/1f5630af51f24d79053b6bef5b8b3ba93d6373... rtmp_packet_read_one_chunk] function. Size of each packet is being read using [https://github.com/FFmpeg/FFmpeg/blob/1f5630af51f24d79053b6bef5b8b3ba93d6373... AV_RB24] each time the packet has been received. If packet with such channel_id does not exist it creates new one using [https://github.com/FFmpeg/FFmpeg/blob/1f5630af51f24d79053b6bef5b8b3ba93d6373... ff_rtmp_packet_create]. Inside ff_rtmp_packet_create space gets allocated using [https://github.com/FFmpeg/FFmpeg/blob/1f5630af51f24d79053b6bef5b8b3ba93d6373... av_realloc] using size variable. Later [https://github.com/FFmpeg/FFmpeg/blob/1f5630af51f24d79053b6bef5b8b3ba93d6373... toread] value is computed which really means {{{toread = FFMIN(value, 0x80) }}}) and passed to [https://github.com/FFmpeg/FFmpeg/blob/1f5630af51f24d79053b6bef5b8b3ba93d6373... ffurl_read_complete] function. If [https://github.com/FFmpeg/FFmpeg/blob/1f5630af51f24d79053b6bef5b8b3ba93d6373... size] value is still more than zero we have to read once [https://github.com/FFmpeg/FFmpeg/blob/1f5630af51f24d79053b6bef5b8b3ba93d6373... more]. Next time we can send another size value and it is not being checked that is different from the previous one, so more bytes have been read than size of the current packet. '''Steps to reproduce:''' I am attaching PoC file that reproduces the simple crash. Steps to reproduce bug: {{{ $ python rtmp_poc.py & $ ffmpeg -v trace -i rtmp://localhost:12345/ $ gdb -q ./ffmpeg_g (gdb) r -v trace -i rtmp://localhost:12345/ }}} I am also attaching gdb stack-trace. '''Recommended fix:''' Checking that the size has not been changed from one chunk to another would do it. '''Final words:''' Our team plans to release full exploit and blogpost with full details on exploitation in 30 days or as soon as you patch vulnerability. Contact me if you need more details on vulnerability. -- Ticket URL: <https://trac.ffmpeg.org/ticket/5994> FFmpeg <https://ffmpeg.org> FFmpeg issue tracker