DM6467T vertical sync loss on Blue Ray playback

Hi,

Do you see any VPIF error interrupt when this problem occurs? If VPIF gives error interrupt in this case, this error could be notified to the application.

If VPIF does not give error interrupt, it is difficult to handle. This is because as far as i remember, tvp7002 does not have any lock status register, so it is difficult to know whether lock is lost or not.

Thx,

Brijesh

Brijesh,

Thanks for looking through this inquiry.  There are no traces from VPIF and nothing in dmesg after "vpif_display vpif_display: Pixel details: Width = 1280,Height = 720" which is traced when the display starts up.  I am going to enable debug traces on the VPIF capture driver in case the loss is not being emitted.

If VPIF does not observe the sync loss, what other options do we have?

Thanx,

-Tomas

Hi Tomas,

I think this error is not registered, so dmesg will not provide any log for this error interrupt. You can check this interrupt in vpif_set_intr register. If error interrupt is enabled, 5th in this register should be enabled. 

If VPIF does not detect any loss of sync, then we have to detect at the source/decoder level only. See if there is any lock bit in the TVP, which can be polled after every fixed amount time. I am sure that lock bit is available in TVP5147, but not sure about TVP7002.

Thx,

Brijesh

Brijesh,

Ok.  That gives us the necessary confidence to make changes to the VPIF driver.  We can work around the loss of vertial sync, but I would really like to confirm that we are not faced with a different type of issue.  Is it unusual for video loopback to loose sync on Blue Ray playback?  Our cables are short (6ft) and other devices don't exhibit this flaw during the playback of this disc.  Even if we do respond to the VPIF interrupt, we will need to STOP and START the stream which will drop frames.  If there is a problem with the analog hardware or TVP7002, it would be evident in higher resolutions.

Our understanding is that one of the benefits of the 6467T (as opposed to the DM6467) is that is can support the 1080P input pixel clock rate, however, when I switch to 1080P, the TVP7002 cannot determine the video standard.

It's strange because the first PS3 screen that waits for me to confirm the new video mode is detected as video standard 17, but the frames scroll vertically.  As soon as I confirm, the PS3 adjusts, and subsequent calls to Capture_Opens yield that the input video standard is unknown.  The trace is:

@0x00080c74:[T:0x4001df90] ti.sdo.dmai - [Capture] Component input selected
@0x00080ede:[T:0x4001df90] ti.sdo.dmai - [Capture] Checking video standard
tvp7002 1-005d: querystd, erorxxx, val = 1, val1 = 3d2
@0x000bb14d:[T:0x4001df90] ti.sdo.dmai - [Capture] VIDIOC_QUERYSTD failed on /dev/video0. Video input connected?

The val1 ranges from 0x0 to 0x464.  This value is read from the TVP registers that code the number of scan lines per frame.  Since the values vary every call to ioctl(QUERYSTD, ...), I think it's clear that we cannot vertically sync the signal.

Has anyone observed variation between boards as far as video content loopback as we observe w/ 720P?  Can the 6467T support the 1080P input?

Thanx,

-Tomas

Hi Tomas,

I also used Sony PS3 for 1080p@60 testing and i think PS3 supports only blue ray discs. I never faced any issues with Blue Ray discs. Capture always detected correct mode when QUERYSTD ioctl is called. I can two differences between my setup and yours. The first is since i did not have disc with the 1080p content, i tired with some already stored content. I do not think this could create any issue. The second is i first changed mode on PS3, test it by directly connecting PS3 to TV and then, i connect PS3 to DM6467T. Could you try in this sequence?

Are you seeing split image on the output when VPIF looses sync? If it is the case, TVP has lost the sync and since VPIF DMA is on, they both have gone out of sync. The easiest way to make them in sync is to reset VPIF. You dont need to restart the streaming in the application. When you get VPIF interrupt, i think you can reset VPIF in the driver. There will be a loss of few frames because of resyncing.

Thx,

Brijesh Jadav

Brijesh,

Yes.  We are seeing the split image on output.  I tried the sequence you recommended.  I have connected the PS3 to a standard display and then to the EVM.  We still have the same problem.  The disc we have is described as 1080p HP widescreen 1.78:1 feature, however, this shouldn't matter as QUERYSTD still does not work - even when we are in the PS3 menu.

We are using the arago linux-davinci-staging branch for the kernel.  We are also using arago for the rootfs and dmai build.

I will try resetting the VPIF capture, but it doesn't seem like we should have to do this.  Can you help us determine that our hardware platform is valid?  It sounds like you have no problem setting the input format to 1080p.  I'm assuming this is configured under your Display -> Video Settings.  Can you confirm?

If so, maybe there is an issue w/ our EVM.. ?

Thanx,

-Tomas

Hi Tomas/Brijesh,

Did you managed to resolve this sync loss? I am also facing the same problem. I am using ADV7441 with DM6467/T and able to detect sync loss by reading ADV lock register. Once I detect the sync loss I reset the VPIF  but no luck.

Its not like the solution does not work all the time. I see sync loss issue time to time.

--

Regards,

Krunal

Hi Krunal,

Unfortunately, we were developing an HD video broadcasting prototype for a prospective client and had to cut our losses with this issue.  It turned out to be highly dependent on the input media, likely due to VPIF vertical sync loss due to a glitch in the media. (inappropriately sized frame for the resolution, or overscan)  We certainly have the capability to resolve this issue internally, likely with more robust handling of the VPIF error state in the kernel driver, however; we need a paying client.

Different chapters and blue-ray DVDs did not exhibit this issue so we could simply avoid that particular scene.  It is likely resolved in TI's updated SDK/Linux kernel driver, but we haven't had an opportunity to bring that platform back up.

Wish I could be more help, but check the vpif driver updates at arago-project.org to see if TI posted improvements.

Thanks,

-Tomas