[CrazyCoder]

The TV is set to Unscaled mode and the input is set to PC mode. When I connect a Windows PC via HDMI to the same TV input, there is no overscan issue. PS3/PS4/XBOX and Chromecast connected to the same input (via receiver) also do not have overscan issue.

The issue is specific to Chromebox device which has no HDMI output and is connected via DisplayPort to HDMI cable. Not sure if the issue is caused by the cable.

ChromeOS also has the same problem, so it's not specific to Ubuntu on this device, rather to the device itself, the driver or the cable.

Autodetected ModeLine: http://pastebin.com/Z3FE9ZHL .

[CrazyCoder]

xrandr output: http://pastebin.com/NZmyK7tA
I've switched to 1080i using "xrandr --output HDMI3 --mode 1920x1080i" command and there is no overscan in this mode, the issue is specific to the first mode (1080p).

[fritsch]

xrandr --verbose

the interlaced mode sucks. Better get a better adapter

[CrazyCoder]

I've connected via DVI port instead of DisplayPort using HDMI to DVI adapter, it has the same overscan issue as with DisplayPort connection.

Here is the verbose output for the DVI connection: http://pastebin.com/ysDsawUw .

[CrazyCoder]

Wow, it was actually a TV firmware bug fixed just a week ago:

QF2EU-0.173.46.0 (Date published: 2014-02-10)

• Solution for picture moves 5 pixel right when in PC/ Game mode.

I have no idea why this issue didn't appear when a Windows PC was connected in the same mode.
Just updated the firmware and it resolved the problem with Chromebox. Sorry for the unrelated flood.

BTW, I've compared the verbose xrandr output before and after firmware update and the only difference was CRTC being 0 before the update and becoming 1 after the fix.

[fritsch]

https://github.com/FernetMenta/xbmc/comm...987fb9f739 <- hevc now multithreaded.

[noggin]

Wow, it was actually a TV firmware bug fixed just a week ago:

QF2EU-0.173.46.0 (Date published: 2014-02-10)

• Solution for picture moves 5 pixel right when in PC/ Game mode.

I have no idea why this issue didn't appear when a Windows PC was connected in the same mode.
Just updated the firmware and it resolved the problem with Chromebox. Sorry for the unrelated flood.

BTW, I've compared the verbose xrandr output before and after firmware update and the only difference was CRTC being 0 before the update and becoming 1 after the fix.

It could have been that some devices were sending RGB via HDMI and others were YCrCb via HDMI and the pixel shift only happened with one format not the other? Just a thought - but it seems irrelevant now that your TV has been updated to fix the problem.

[RipT]

I had another crash watching the Olympics, but this time it is reasonably reproducible. It always happens when doing a step-forward during the ice dancing. It's not always in the same place, but close. I suppose it's possible that it's some corruption in the video file from poor TV reception, but it wasn't anything obvious watching the video.

I have mesa 10.0.3 running now, and I did confirm that I have gdb installed, but it was apparently installed before the last set of crashes, so I'm not sure why there aren’t any obvious debugger messages if the build was configured for them.

Anyway, Since it was reproducible, I was able to turn on the verbose debug logging. Here's the Crashlog:

http://paste.ubuntu.com/6947468/

I'll try running the video in a new user with no extra addons to see if that makes a difference. Let me know if you see anything informative or have any tests you think i should try.

Thanks for the help and for making this branch. it really does look a lot better than the mainline on interlaced content.

[fritsch]

sudo apt-get install gdb

and recrash

From what I see, blame myth:

23:55:35 T:140274640054144 INFO: AddOnLog: MythTV cmyth PVR Client: Received announcement: Player, xbmc, OnSeek

[RipT]

OK, it didn't take as long as thought to get it running in a new user. I couldn't get it to crash until I enabled software filters and set it to "deinterlace" (yadif) for the de-interlacing method. Since it doesn't always happen in exactly the same spot, this could be a false correlation (in fact I couldn't reproduce it this time with the verbose debug logging activated), but I did get it to crash with the "deinterlace" setting, and nothing else activated (including no PVR). I notice there is a stack trace in the crashlog, so there's something wonky about the other user I was in before. Here's the new crashlog:

http://paste.ubuntu.com/6947597/

I'll keep trying to get it to crash with the verbose logging. Please let me know if you see anything in here that gives a clue to this.
Thanks again.

[RipT]

OK, I got it to crash with verbose logging. Heres the crashlog:

http://paste.ubuntu.com/6947676/

I actually wish it was the PVR at this point. I wouldn't mind turning off the addon if that was a workaround.
It does seem harder to reproduce with all the addons gone, but it does still happen.

[FernetMenta]

please disable/uninstall version check addon. the python call stack looks not ok. Then post a new crashlog if the issue still exists.

[RipT]

There is another thread over in cmyth that also has a report of crashing when stepping forward. There is a crashlog. It's a recent nightly build running on Ubuntu 12.04, but not this branch. The thread is http://forum.xbmc.org/showthread.php?tid=186324.

[fritsch]

Here is a sample, just loop them:

1080i50: https://dl.dropboxusercontent.com/u/5572...g-test.mp4
576i: https://dl.dropboxusercontent.com/u/5572...formers.ts

[RipT]

I disabled Version Check, but it still crashed. Here's the log:

http://paste.ubuntu.com/6950771/

@fritsch - Did you want me to run those sample videos to try to generate a crash? The one logged here was stepping through and Olympics broadcast like the previous ones.