[mythtv-users] HD-PVR Priming Script?

[Martin Lynch]

>> Aside from my ongoing RBlaster woes, my new HD-PVR routinely (read: at
>> least twice a day during heavy use) fails. By this I mean that stopping
>> watching live TV in myth doesn't stop the HD-PVR - the blue light on top
>> remains on, and myth can no longer use that input thereafter. Likewise,
>> live TV fails a couple of times a day with an error message along the
>> lines of "video failed - too many buffer overruns". The HD-PVR stays
>> lit. If I manually turn the HD-PVR off and then on, myth often doesn't
>> recognize it - the backend gives an error that there are no inputs
>> attached. As near as I can figure out, my entire system requires a
>> reboot at this point in order for the HD-PVR to work (ie I've tried
>> removing the capture card in mythtv-setup and replacing it, which does
>> not work, etc.).
>>
>> What I'm wondering is: would it be possible to write a script which
>> tests the HD-PVR and, on a bad result, power cycles it or, if necessary,
>> restarts the machine after a suitable warning and delay? Something like:
>> a minute prior to a scheduled recording, attempt a 2 second recording
>> and then check the file size. If>0, OK else power cycle the unit. I
>> have no idea how to actually do this; nor do I know whether it is
>> possible to trigger a power cycle or reset of the HD-PVR via
>> instructions over the USB cable. I just got to thinking about this
>> because I've used firewire for four years with excellent results, and if
>> I recall, the firewire capture code includes a priming script that makes
>> sure a solid connection is established, and I thought something similar
>> might be possible with the HD-PVR.

> I don't know if it's definitely the case with the HD-PVR, but it's 
> possible that the reason Myth fails to recognize the device after it 
> is reset is that it is treated like a second instance. This would 
> create a new /dev/videoX device. If this is the case, then you might 
> be able to work around the problem with a udev rule that creates a 
> symlink with a constant name to whatever device gets created. You 
> would then point Myth to the symlink instead of the videoX device. You 
> could confirm this using `ls /dev/video*` before and after resetting 
> the HD-PVR. It might also help if you "rmmod" the driver after 
> unplugging the HD-PVR.
> -- Dan Wilga "Ook."
Interesting - was hoping to try this with today's first failure, but 
instead the failure froze the entire machine, requiring a hard reboot 
(ie I was unable to even ssh into it)...

-- 

Kind regards,

Martin