[pvrusb2] Mythtv hangs on livetv

[Julio Arruda]

Mike Isely wrote:
> On Tue, 13 Dec 2005, Julio Arruda wrote:
> 
>> From a couple of tests I did few weeks ago, is not exactly MythTV 
>> hanging..
>> It seems when you start to 'read' from the livetv buffer (or from the 
>> blablabal .nuv 'non-live' recordings in my case also, while they are 
>> being recorded),the pvrusb2 driver can't read from the PVRUS2 box 
>> anymore.
> 
> 
> Are you seeing any log messages from the driver which might suggest this?
> 
> If you go back to an older driver snapshot does the problem still happen?
> 
> If you cat /dev/video0 (or whatever your /dev name is), does it fail there?
> 
> If you try mplayer /dev/video0 (or whatever your /dev name is), does it 
> also fail there?

Can't help a lot right now on this, had to remove all my gear from the 
"production" mythtv box..I'll need to reinstall the PVRUSB2, but in 
fact, the combination I had was a little more complex (video0 is an 
existing pvr150 running in the same box). I have another box where 
(having the time) I should be able to install the pvrusb2 box (sadly 
with USB1.1 in this case) and test it there.

> 
> Diagnosing a problem like this is a matter of searching out clues and 
> ruling out common causes.  Though the pvrusb2 driver has had some recent 
> changes, they are largely cosmetic in nature, so I would expect bugs 
> introduced by such changes to NOT be subtle.  (The buffering code in 
> particular has not changed at all.)  On the other hand I have learned 
> over time that MythTV is an unending source of stability headaches - 
> stuff that works one week stops working in a future snapshot.  Somebody 
> here recently said that he's using a recent snapshot because "now it's 
> possible to watch a recording while it's being recorded" - well I've 
> been doing that with version 0.18.1 since last May so I guess this was 
> something that had been broken and since reintroduced as a new feature.

This 'watching a recording while it's being recorded' in fact would be a 
good description of what I saw, looking like the trigger, but I'll need 
sometime to try it again.

> Anyway, rather than laying blame here, what really needs to happen is 
> that we have to rule things out.  If for example somebody can 
> demonstrate the hang with mplayer, then MythTV will have been ruled 
> out.  On the other hand if mplayer continues to be stable, if you can 
> "cat" the video for hours on end, and if xawtv is stable (well, at least 
> if it is as stable as it usually is), then that might point the finger 
> at MythTV.  Along with that, if you can extract suspicious entries in 
> your kernel log from the pvrusb2 driver (like for example the driver 
> complaining about timeouts to the hardware), then that's a solid clue 
> suggesting a driver and/or a hardware problem.
> 
> The only clues I see so far are in the log that Roger sent, and it is 
> pretty suggestive of some kind of clash between MythTV and the X server 
> he is running.
> 
> I'll keep an eye out for problems on my end, but so far I haven't seen 
> any (well I haven't seen any *new* problems).
> 
> We need more information here.

Sure, I agree 100%, reproduce and isolate the problem to the 'simpler 
config where it will happen' should help :-), I will try to get it done 
during the next 'maintenance window' here :-)
BTW, amazing work in the pvrusb2 driver, it is just the perfect matching 
to a small form factor pvr with several tuners :-).