[pvrusb2] xawtv fix

[Andreas Beck]

Mike Isely <isely at isely.net> wrote:
> On Wed, 12 Apr 2006, Andreas Beck wrote:
> > It does for me. I changed
> > ret |= pvr2_write_encoder_vcmd(hdw, CX2341X_ENC_SET_OUTPUT_PORT, 2,
> > 								0x01, 0x02);
> > to 								0x01, 0x00);
> > and it still works.
> That's exactly what I was looking for!  Excellent.
> > I have not checked, however, if it would now work with an unpatched
> > xawtv.
> Please try this with an unpatched xawtv.  This is significant.

Done and still works for me. If it works with the old HW as well, 
this should be nailed down, then.

> > Moreover, I can confirm audio working for me now, when using the TV
> > tuner.
> > However, it still does not work, if I use the composite inputs.
> Ah, I must have missed that before.  For the composite audio input, the 
> audio pathway travels through the wm8775 chip on the device (before 
> reaching the cx25840), 

Ahhh ... I have just tried all inputs listed in the cx25840 kernel
header - but to no avail.

I have learned how to sample the composite input with audio from the tuner
now, but this is of very limited use ...

I was under the impression that the audio inputs might have been wired 
in another way now, but your post gives quite some insight. 

I was already wondering, why there was a "SERIAL" input which seemed to
be so different from the other inputs ...

> and that part is not used when using the tuner. The wm8775 is an 
> analog->digital converter for the external analog inputs 
> (when using the tuner, then cx25840 handles that job directly). 

I see. I didn't think of such a design ...

> So the problem is clearly with wm8775.  Check that wm8775.ko is in 
> your kernel 

it is

> and that it is attaching to the driver (cat 
> /sys/class/pvrusb2/sn-XXXXXX/debuginfo to see the list of attached 
> modules). 

it is not. 

> There is a special step you have to take right now to get 
> wm8775 to attach, because right now the chip isn't being detected 
> correctly.  Reload wm8775.ko with the option "force=-1,27".  That will 
> bypass the detection.

Now it is listed as attached.

STRIKE. That did the trick. Thanks a bunch.

(BTW - I also get the audio distortion, but I got to recheck, as the
 laptop I am using has really crappy speakers, so it might be just that.
 Hmm - seems to be overshooting on the machine with reasonable speakers
 as well. I haven't tested with 0x02 for the second parm yet.
 If this keeps being an issue, I could borrow a reasonable sine 
 synthesizer to check distortion on an objective scale.)
 

CU, Andy

-- 
= Andreas Beck                    |  Email :  <becka-sig at bedatec.de>         =