On Thu, Jun 16, 2005 at 11:21:08AM -0400, Isaac Richards wrote: > On Thursday 16 June 2005 09:47 am, Robert Tsai wrote: > > I am going to blindly hypothesize that this might be related to > > the AMD64-specific pthreads problem for which exists Kyle's > > workaround in RingBuffer, but there don't appear to be any > > readwrite locks involved in this bit of the code (although I'm not > > familiar with QMutex). > > Actually, it's quite possibly to be the same issue, if mythweb is > requesting the backend to generate thumbnails for the recordings. > That will involve the RingBuffer class (to read the local video > file), and if enough of those queries stack up due to problems > there, you'd see the current behaviour.
But I already have Kyle's RingBuffer patch applied, so if it's just re-using RingBuffer code, then it shouldn't be this exactly same issue, right? > It shouldn't be terribly hard to get a backtrace of the backend in > its stuck state - it'd be interesting to see where the threads are. Anyway, yeah, I'll debug later. I don't know about getting an "all threads" backtrace, though. I tried that, and I had some threads that had call stacks of 1000 frames and counting (had to kill gdb). --Rob
signature.asc
Description: Digital signature
_______________________________________________ mythtv-dev mailing list [email protected] http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-dev
