[pvrusb2] mythtv and wintv-hvr-1950 - analog/digital scanning

[Mike Isely]

On Tue, 6 Jan 2009, fivenote wrote:

> Great tips. Here's what I found...
> 
> I disabled mythtvbackend and used sysfs interface to control the
> driver. I also confirmed it is creating /dev/video0.
> 
> The enum_val shows all 4 inputs (dtv, television, composite, svideo).
> For fun I restarted mythbackend and mythtv and watched enum_val narrow
> successfully between dtv and television. DTV signal came through, but
> still no television signal came through.
> 
> I disabled mythtvbackend again and checked sysfs. All four inputs
> showed, meaning nothing was left open and narrowed.

OK...


> 
> I switched cur_val to 'composite' and hooked up a video player to the
> HVR-1950's composite input. mplayer still didn't work. I tried xine
> and it says "There is no demuxer plugin available to handle
> '/dev/video0'. I'll try to figure out what that is. Any hints are
> welcome.

Did you get the same failure as before with mplayer?  Or is it different 
now?

But forget mplayer.  The *easiest* thing - by far - to try is just to 
cat the device node, as you tried before.  That should always do 
something noteworthy.  You said you tried that before and got an EIO 
("input/output error").  I am 99% certain you got that EIO because 
something else was streaming from the device at that time.  There is 
very little other reason for this occur.  Inability to tune does *not* 
result in an EIO.  Ever.  You should get something.  So, given that 
you've confirmed now that you've cleared everything else out of the way, 
does "cat /dev/video0 >/tmp/foo.mpg" still result in an EIO?  It should 
not.

As for xine, the error you report sounds a lot like xine tried to open 
/dev/video0 as a DVB node, which is a wrong thing to do.  The node 
/dev/video0 is the V4L interface and attempts to open that with an app 
that tries to treat it as a DVB interface will only result in failure.  
The DVB interface path is entirely different; it is part of the DVB 
interface standard ("/dev/dvb/adapter0/....").

If you try to 'cat /dev/vide0 >/tmp/foo.mpg' - given that NOTHING else 
is trying to stream from the hardware - and you still get an EIO, then 
that's noteworthy.  More likely the cat command will do something more 
useful, either result in a file full of video snow or a black screen, or 
it will just block generating no data (which you can interrupt), or it 
will fail with some other error.  (Or it might even work.)  Knowing the 
results of this is a big clue.  Definitely try this with the input 
selection via sysfs set to composite.


> 
> Meanwhile... fyi... When I installed v4l drivers, it had it's own
> pvrusb source and replaced the one I downloaded from Mike's site. Is
> there a difference between the pvrusb2 source at Mike's site and
> LinuxTV? Which is more current. I figure this is worth mentioning so
> we know what source I'm running.

They are both completely current.  The differences (when comparing the 
latest versions of each) are compatibility:

The standalone driver has a lot of extra bits and pieces that allow it 
to compile and work correctly for all known Linux kernels back to 
2.6.12.x and all known corresponding versions of v4l-dvb (i.e. those 
that work with those kernels).

The in-v4l driver is trimmed to contain only that which is relevant to 
the specific surrounding code in the v4l-dvb snapshot.

The in-kernel driver is further trimmed to contain only that which makes 
sense for the specific kernel version it was bundled with.

Feature/function/stability wise, all three are equivalent.


> 
> Thanks for all your help. I'm learning more each time and hope to
> figure it out soon.

You are trying to debug a confusing problem.  First step to solving the 
issue is to remove the confusion and simplify things down.  When 
testing, only change one variable at a time.  Be methodical about what 
you try - don't just do random combinations of things but try simple 
things first and build up from there as you verify the setup.  For this 
reason, FORGET MythTV right now.  And don't bother with mplayer (yet).

Right now you don't even have the simplest case working, so focus on 
that first before dragging in everything else.  Use the sysfs interface 
to control the device and use cat to stream from /dev/video0.  Stick to 
the composite input and forget the RF tuner.  Make sure you can 'cat 
/dev/video0 >/tmp/foo.mpg' and that you get a non-zero file result.  
Then examine that /tmp/foo.mpg file with mplayer.  If mplayer then shows 
you have valid video, only THEN do you introduce other factors.  For 
example, I'd probably then try running mplayer directly on /dev/video0, 
and if that now fails re-read the pvrusb2 driver's web page notes about 
mplayer and make sure you are using it correctly - because if cat works 
then there's nothing wrong with the driver that should keep mplayer from 
working.

Once you've got mplayer working with /dev/video0, then I'd turn my 
attention to the RF tuner and see if you can receive a broadcast.  
(Note: US citizens using OTA broadcasts only get until roughly middle of 
this February to do that test...)  Once you verify that then you've 
proven that the entire analog side of the device is healthy.

Next I'd turn my attention to the DVB side.  You had that working with 
MythTV before so it's probably fine.  But for the record, one test I've 
done on the DVB side is to download and build the dvb-apps repository 
and use the azap utility to open the DVB interface and lock a station 
and then use mplayer to stream via the corresponding DVB interface.  
(This technique is not specific to the pvrusb2 driver; any utility / 
trick you can do with a DVB device will work here.  So just try 
something simple.)

Remember, we're not throwing random stuff at a wall to see what sticks.  
You need to build a picture of what is going wrong, and you have to 
build it up starting with simple things, adding more factors as the 
simpler stuff is verified.  These are good things to remember any time 
you're trying to diagnose a problem anywhere, not just with this driver.

Hope that helps.

  -Mike

-- 

Mike Isely
isely @ pobox (dot) com
PGP: 03 54 43 4D 75 E5 CC 92 71 16 01 E2 B5 F5 C1 E8