[mythtv] null tvchain in TVRec::TuningNewRecorder

[George Nassas]

Hi all.

Over the weekend I upgraded from a mid-November SVN to recent code and 
now I can't make recordings anymore.

I've been poking around the code all afternoon and the problem seems to 
be when a recording starts there's a null value in TVRec.tvchain and 
this causes TVRec::TuningNewRecorder to skip the ringbuffer setup code 
(lines 3641 to 3657 in svn 8870). Following that much mayhem ensues. On 
the other hand, if I'm in live TV when it's time to start a scheduled 
recording then things work OK.

By way of background I'm running svn 8891 on a debianized kernel 2.6.15 
and ivtv 4.2. I did my upgrading in two steps: the kernel and ivtv were 
upgraded from kernel 2.6.8 and ivtv 3.X on Thursday and ran them 
against the mid-November myth for two days. Things seemed stable so I 
moved myth to current svn and haven't had a successful recording since. 
Oh, there's also an xorg upgrade in there, I went from 6.8.heinz57 to 
6.9 if that matters.

Oddly, if I reboot with the 2.6.8/3..X combination I'm able to make 
recordings although they're poor quality so I'd really rather not go 
back to the older driver.

I'm continuing to analyze this but I wanted to open it up to people who 
are more familiar with this code than myself. There may be an 
environmental component to this and I admit my machine is a bit on the 
crufty side but still, that TuningNewRecorder method really ought to 
die violently if it doesn't have a recorder to tune.

Thoughts? Advice?

- George