Comment # 6 on bug 941867 from Takashi Iwai

(In reply to John Johansen from comment #4)
> Created attachment 649832 [details]
> Temporary workaround patch
> 
> I have attached a patch that will keep the bug() call from triggering. It
> does not address the root cause, as it leaks some memory instead of dealing
> with why the subprofiles have not been freed at this point.

Yeah, BUG() seems like a overkill for small memory leaks.  Maybe WARN()?

I'm going to backport this in anyway.  Meanwhile, would you merge this to
upstream, too?