Hello, My backtraces never actually mention play_sound, but the crashes only happen right after app_conference attempts to play out DTMF tines with the playing function.
Here's the backtrace for two of the crashes that we had with app_conference: http://205.201.151.24/files/app_conference-crash-2006-06-02.txt http://205.201.151.24/files/app_conference-crash-2006-06-05.txt MATT--- On 7/12/06, jeff oconnell <[EMAIL PROTECTED]> wrote:
interesting. i didn't realize the problem seems to specifically be the sound playback via the manager interface. a couple weeks ago, my asterisk on my dev box crashed, i did some preliminary investigation, but since we hadn't had any problems in production or qa, i chalked it up to me messing up my dev install somehow. anyway, here's the backtrace of the core: (gdb) bt #0 0x00df71bb in ?? () from /lib/libgcc_s.so.1 #1 0x080625ad in ast_deactivate_generator (chan=0x815fb40) at channel.c:1382 #2 0x0806d77a in ast_openstream_full (chan=0x815fb40, filename=0x815dcec "tones-exit", preflang=0x0, asis=0) at file.c:494 #3 0x0806d835 in ast_openstream (chan=0x71bfb8a6, filename=0x71bfb8a6 <Address 0x71bfb8a6 out of bounds>, preflang=0x71bfb8a6 <Address 0x71bfb8a6 out of bounds>) at file.c:467 #4 0xb7e8899b in conference_play_sound (fd=12, argc=14643636, argv=0xdfdc5c) at cli.c:225 #5 0x0809786b in ast_cli_command (fd=12, s=0x71bfb8a6 <Address 0x71bfb8a6 out of bounds>) at cli.c:1364 #6 0x080b2f8a in action_command (s=0x815c448, m=0xb7b97420) at manager.c:927 #7 0x080b7b81 in process_message (s=0x815c448, m=0xb7b97420) at manager.c:1305 #8 0x080b83cf in session_do (data=0x815c448) at manager.c:1401 #9 0x00b44341 in start_thread () from /lib/tls/libpthread.so.0 #10 0x009096fe in clone () from /lib/tls/libc.so.6 the interesting lines to me are #4 and #5: #4 0xb7e8899b in conference_play_sound (fd=12, argc=14643636, argv=0xdfdc5c) at cli.c:225 #5 0x0809786b in ast_cli_command (fd=12, s=0x71bfb8a6 <Address 0x71bfb8a6 out of bounds>) at cli.c:1364 line 4 becauset the argc passed into conference_play_sound() is so large, and line 5 because there seems to be an out--of-bounds problem in the asterisk code ( i.e. before the app_conference code is called ). based on what you said in your last post, i'm going to look at this more. if you have any thoughts on my backtrace/analysis, let me know. j- On 7/12/06, Matt Florell <[EMAIL PROTECTED]> wrote: > The issues with the double-free crashes that we've had all seem to be > caused by playing of audio files(like the entry/exit sounds or the > DTMF broadcast). But these functions use the app_conference code that > was already existing to play audio files from manager API commands so > the issue was there it just wasn't as tested because not many people > use the manager command a lot to play audio in conferences. >
_______________________________________________ --Bandwidth and Colocation provided by Easynews.com -- asterisk-users mailing list To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users
