DM6467 VPIF Capture bug

I am afraid, re-synchronization is outside the scope of VPIF specification; therefore this is more of a hadrware limitation than a bug in the software driver. 

That said, we have seen some users uses the I2C Sync Detect Status register on TVP5147 or TVP7002 to implement a polling mecahnism and detect when sych has been lost (disable VPIF) and when synch has recovered (enable VPIF) to work-around this issue.

look at the following App Note for a workaround: http://focus.ti.com/general/docs/litabsmultiplefilelist.tsp?literatureNumber=sprab77

I tried the sprab77 workaround with the DM6467, but it does not help - image is still split after a disconnect-connect sequence.

Is there anything else to be done besides disabling/enabling VPIF?

Are you resetting VPIF after reconnecting the source?

No, I follow sprab77 document - according to the example in this document once the sync is lost, the LSB bit in VPIF_CHO_CTRL register is set to 0 and as soon as the sync is detected, the same bit is set to 1.

Anyway, I also tried resetting after reconnect (i.e. when sync is lost, just wait for detecting it again and after detecting it, set the bit to 0 and then set it to 1), but this didn't help and I still get the split image.

Found the problem.

TI's solution works OK. The problem was that I use different video decoder from the one on the EVM and the registers which I checked in this decoder tell only if video signal is present on the input but cannot tell if the decoder actually outputs video data. Intorducing a delay after checking these registers (as a workaround) and then setting the LSB of VPIF_CH0_CTRL to 1 solves the issue.

FYI, sprab77 worked for me as well.  I was taking two NTSC inputs from two TVP5147's.  I wrote a combo function that checked both/either with the same loop.  Also, note that I'm using dvsdk_3_10_00_11 and the code examples in sprab77 were TOTALLY INVALID, at least it seemed that way to me.  Nevertheless, with the implied advice I was able to figure out how to do it.  I modified the TVP5147 driver to have a function tvp514x_checkNTSClock that checked strictly for the lock on an NTSC signal.  The caller to this function [messily] used a global that I made tvp514x_probe set, for the address of the v4l2_subdev, based on matched I2C address.  The caller was board-dm646x-evm.c where I wrote a function that was called from an ioctl I had developed earlier.

Juan Gonzales said:

look at the following App Note for a workaround: http://focus.ti.com/general/docs/litabsmultiplefilelist.tsp?literatureNumber=sprab77

This post is old, but it seems people are still experiencing this issue.  I wanted to post the link again, since as Mike stated, that link it broken. 

http://www.ti.com/lit/an/sprab77a/sprab77a.pdf

Thanks,

Derek