Comment 174 for bug 405294

I'm generally fine with getting this approved, but before that I'd like some more regression potential analysis to be done and included in the Regression Potential field. The first patch added to fix the issue is quite big and includes a rewrite of thread_func() in there - and any rewrite with so much code modified carries some regression potential. What we'd like to know is which parts of pulseaudio's functionality could regress in worst case scenarios, at which parts testers should look to expect issues.