[pvrusb2] cx25840 video scaler is broken [was: Poor analog picture on HVR-1950 after upgrade...]

[fivenote]

Congratulations Mike!

Your description of how the device initialization works makes total
sense with what I observe. I always noticed that the HVR-1950 starts
up in digital mode. Also, sometimes the switch to analog doesn't work
smoothly. I sometimes need to restart the driver or re-power the
device before the digital->analog switch works. I'm sure your work on
the video standard switching will continue to improve things.

Thanks again.

-vincent

On Sun, Jun 14, 2009 at 1:30 AM, Mike Isely<isely at isely.net> wrote:
>
> Problem found...
>
> The VBI hack wasn't working for 24xxx in v4l-dvb w/ sub-device binding
> because I was calling it too soon.  The driver was setting up the VBI
> hack, THEN loading the cx25840 firmware and initializing the device
> which of course is all backwards.  I've just tested a minor change to
> fix this ordering and sure enough 640 is working correctly again.
>
> Knowing that, I had an idea about the HVR-1950 scaling problem...  It's
> a different but similar problem.  In that case the pvrusb2 driver is
> issuing a reset to the cx25840 driver whenever it switches the hardware
> to analog mode.  I strongly suspect that this reset action is going to
> erase chip state, including whatever the VBI hack did.  And even if you
> don't run the device in digital mode this still happens - because the
> pvrusb2 driver will infer a switch to digital mode at device
> initialization in order that the DVB interface can initialize.  Thus the
> cx25840 hardware is NEVER correct there.  This is only a theory; I
> haven't tested a fix here yet (which would be to ensure the chip is
> correctly reinitialized after a reset).  But I'm 90% sure that this will
> fix analog mode when scaling is active in the cx25840 for HVR-1950
> hardware.
>
> Actually, looking at the VBI setup within the cx25840 driver code, it
> looks like I really should be re-issuing it any time the video standard
> changes.  In fact I think it's still being run too soon because the
> driver is being told the video standard *after* this hack is done.  So I
> need to take a step back and examine this situation a little more
> carefully.  But it's just a matter of time - I'm pretty sure I've got
> this problem under control now.
>
> Whee!!!
>
>  -Mike
>
>
> --
>
> Mike Isely
> isely @ isely (dot) net
> PGP: 03 54 43 4D 75 E5 CC 92 71 16 01 E2 B5 F5 C1 E8
> _______________________________________________
> pvrusb2 mailing list
> pvrusb2 at isely.net
> http://www.isely.net/cgi-bin/mailman/listinfo/pvrusb2
>



-- 
~~~~~~~~~~~~~~~~~~~~
VAO - fivenote at gmail.com
~~~~~~~~~~~~~~~~~~~~