[mythtv] [mythtv-commits] Ticket #1805: Backend not responding, with backtrace and log

[Jim Westfall]

MythTV <mythtv at cvs.mythtv.org> wrote [05.13.06]:
> #1805: Backend not responding, with backtrace and log
> -------------------------------------+--------------------------------------
>  Reporter:  tom at redpepperracing.com  |       Owner:  ijr
>      Type:  defect                   |      Status:  new
>  Priority:  minor                    |   Milestone:     
> Component:  mythtv                   |     Version:     
>  Severity:  medium                   |  
> -------------------------------------+--------------------------------------
>  The symptom that I have been experiencing is that the backend will all of
>  a sudden stop  responding to connection requests. The process is still
>  running, and it's not eating CPU, but neither a frontend nor mythweb can
>  connect to the backend. I finally managed to get a backtrace and backend
>  log when it happened.
> 
>  Version:
>  root at mythtv:/var/log# mythbackend --version
>  Library API version: 0.19.20060331-1
>  Source code version: 9918
>  Options compiled in:
>   linux debug using_lmsensors using_v4l using_oss using_alsa using_ivtv
>  using_lirc using_x11 using_xv using_xrandr using_frontend using_backend
> 
>  Machine is an AMD Sempron 2200, 512MB RAM, 2xPVR-250.
> 
>  See attached backtrace and backend log from around where it happened. I
>  was experimenting with de-interlace and XvMC settings, so I was going in
>  and out of live tv a number of times just before it happened. Not sure if
>  that is related. It also seems to happen around the turn of the hour when
>  a couple of recordings start. Also of note is that the recordings continue
>  to actually record until I stop and restart the backend process, and
>  resume when it comes up.
> 

I assume you have a slave backend too?  

Looking at the backtrace (your master backend?) it appears to have a lock 
on the socket that is waiting for a slave backend to respond about 
generating a pixmap.  At the same time the main app thread is trying to 
process a customEvent, which is blocked trying to get the lock on the 
same socket.  The main app thread being blocked like this is likely 
causing all qt event processing to stop.

jim