[rtmpdump] rtmpserv segfaults when trying to run rtmpdump

[Howard Chu]

blackcode at speedpost.net wrote:
> Attached is strace and gdb backtrace logs.
>
> The rtmp stream is found at
> http://www.bbc.co.uk/news/world-asia-pacific-12725646

strace is useless for crash debugging.

> backtrace.txt
>
>
> GDB Backtrace:
>
> #0  AMF_EncodeInt16 (output=0x0, outend=0xb7710ac4 "\001\024", nVal=8) at amf.c:142
> #1  0x002beee5 in AMF_EncodeNamedString (output=0x0, outend=0xb7710ac4 "\001\024", strName=0x804b4c0, strValue=0x804b4c0) at amf.c:276
> #2  0x0804a217 in SendPlayStart (server=0x804d0d8, r=0xb7710d24, packet=0xb77d5268, offset=1) at rtmpsrv.c:288
> #3  ServeInvoke (server=0x804d0d8, r=0xb7710d24, packet=0xb77d5268, offset=1) at rtmpsrv.c:730
> #4  0x0804a5e0 in ServePacket (server=0x804d0d8, r=0xb7710d24, packet=0xb77d5268) at rtmpsrv.c:822
> #5  0x0804a757 in doServe (server=0x804d0d8, sockfd=9) at rtmpsrv.c:900
> #6  0x0804a8bf in serverThread (arg=0x804d0d8) at rtmpsrv.c:951
> #7  0x0089fe99 in start_thread (arg=0xb77d5b70) at pthread_create.c:301
> #8  0x007ded2e in clone () at ../sysdeps/unix/sysv/linux/i386/clone.S:133

Sort of fixed now in SVN. This stuff is mostly irrelevant so I don't 
particularly care if it's 100% correct. It's just sending the client a fake 
"play starting" status followed by a "play completed" status to try to get the 
client to give up. It apparently doesn't work anyway, the client keeps trying 
to play after the "completed" message.