[mythtv-users] Dying Backend

[Donn]

>it completed recording then died without scheduling 
>the commflag job

I had this problem.  Immediately after a recording ended, backend
would segfault.  This was with a 5/12/05 cvs sync.

I reverted my cvs to 4/22/05 and it has been stable for a week now. 
Another person posted this problem to myth-dev; see subject "SIGSEGV
at end of recording".

Donn
http://mymythtv.blogspot.com

On 5/19/05, Gavin Haslett <gavin at nodecaf.net> wrote:
> I already ran with -v all. I got nothing, the mythbackend process just apparently stopped for no reason. No errors or anything, just log entries showing the last task it completed (it completed recording then died without scheduling the commflag job), then another entry showing that my script had forced a restart.
> 
> I might try the gdb thing as well if I get a chance this weekend. It seems to be a tough problem to recreate on-demand though so it might take a little screwing around to do it. Personally I'm leaning toward a possible hardware issue as it seems feasible. I certainly vaguely remember reading about DMA bugs on VIA chipsets before, and since that's exactly what I'm running and have been encountering what I can only describe as "strange behaviour" when dealing with DMA I'm starting to suspect that's the problem.
> 
> Yesterday I ordered up a new motherboard running a Celeron 1.8 and a half gig of RAM... should be here sometime next week and I can do a motherboardectomy :)
> 
> 
> -----Original Message-----
> From:   mythtv-users-bounces at mythtv.org on behalf of Lucas Meijer
> Sent:   Wed 5/18/2005 6:05 PM
> To:     Discussion about mythtv
> Cc:
> Subject:        Re: [mythtv-users] Dying Backend
> Have any of the people who are having this problem tried running
> mythbackend in gdb, or running mythbackend with -v all option?
> 
> I'm very interesting in your results. If I run mythbackend in gdb, at
> the moment of the crash it says "Cannot find handle 12345: invalid
> thread handle". the -v all debug output shows that the crash is right
> after the scheduler finishes. adding more debugging statements to the
> code shows that it is the fork() call in channelbase.cpp that seems to
> trigger the crash..   a debugmessage before that shows up, a
> debugmessage after that does not show up.
> 
> If you have this problem, could you please try running with -v all and
> in mythbackend and post to this list what happens?
> 
> Thanks, Lucas
> 
> 
> _______________________________________________
> mythtv-users mailing list
> mythtv-users at mythtv.org
> http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users
> 
> 
>