GNOME Bugzilla – Bug 660331
souphttpsrc deadlocks under stress
Last modified: 2014-11-25 18:37:58 UTC
I have a simple app that builds a trivial pipeline souphttpsrc -> fakesink, puts it into a running state, waits for a few seconds, stops the pipeline and exits. This app is run in the loop from two separate machines against custom HTTP server running on another (3rd computer). While stopping the pipeline the test locks up approximately once every two days. The lockup is observed on Linux machine running Fedora Core 15. The core dump is available if necessary. Here is the backtrace of all threads in the application. (gdb) thread apply all backtrace
+ Trace 228616
I did not try to dig deeper, but on the surface it looks like a deadlock caused by souphttpsrc. Regards, -- Max Paklin.
Hi Max, sorry for the long delay in responding. Could you supply the test program by any chance, so we don't have to re-create it ourselves?
Hi Tim, Sorry for the delay, I've been out of town for the holidays. Unfortunately I don't have the sample app that I used to repro the bug. I do remember however that it was simple to write. Really, all it was doing, it was connecting souphttpsrc to fakesink and doing start/stop in the loop. However at this point I don't have either the app or the core dump that I generated back when I was reproing the issue. Sorry for not being helpful. Regards, -- Max.
Thanks for your response. I think it's best to just close this bug for now seeing how much has changed in GStreamer and souphttpsrc since then (0.10 -> 1.x and all). I don't see anyone going to write a test program to let it run for a few days to see if it's still an issue :)