jez250a@ Thanks for the issue.

Request you to provide a sample URL where this issue can be reproduced, which will help in further triaging of the issue.

Thanks..

Hi

Opening various GIFs from this subreddit (https://www.reddit.com/r/SplitDepthGIFS/) will often trigger it. Sometimes it happens almost straight away, sometimes it may take a minute or two.

Another site that often triggers it when there are videos visible on the page: https://www.dailymail.co.uk/home/index.html

Thanks again

Thank you for providing more feedback. Adding the requester to the cc list.

For more details visit https://www.chromium.org/issue-tracking/autotriage - Your friendly Sheriffbot

Hi

I tested the latest 417.71 today with the same problem seen.

Attached GPU error log. The only current workaround I know of is to either revert to 3xx series drivers or to disable hardware-accelerated video decode in Chrome flags.

Thanks

Deleted: ChromeGPU.txt 15.6 KB

ChromeGPU.txt 15.6 KB View Download

Diasbling GPU video decode resolves it, so sounds like it's a problem there. 

The logs look like maybe the gpu video decode code is not recovering from a context loss.

GPU video decode does not recover from context loss. The current expectation is that decoding will fail (DECODE_ERROR), and then playback can be resumed by pressing 'play' again (assuming there is a play button to press in the particular player).

Historically software decode was able to recover by re-uploading frames but this isn't true anymore since the GpuMemoryBuffers are also lost. (However decoding continues and future uploads succeed, so the graphical corruption is temporary in this case.)

We have a mechanism for restarting video decoding now, but have not yet determined whether it should be used to recover from context loss.

The big question here is, 'Why is the context being lost?'

[1556:12172:0115/093146.312:ERROR:gles2_cmd_decoder.cc(16124)] : Offscreen context lost via ARB/EXT_robustness. Reset status = GL_UNKNOWN_CONTEXT_RESET_KHR
[1556:12172:0115/093146.312:ERROR:gles2_cmd_decoder.cc(4583)] : GLES2DecoderImpl: Context reset detected after MakeCurrent.

If i remember, all robustness failures cause the gpu process to exit rather than try to recover.

given the timing, i assume that this is the cause: "testDeviceLost(1850): The D3D11 device was removed, HRESULT: 0x887A0006".

that HRESULT is DXGI_ERROR_DEVICE_HUNG.

This might be a good place to try the d3d11 decoder and see what happens.  unfortunately, there's not an easy way to enable it except via the command line.  i'll add an entry to chrome://flags.  Once that gets to canary, it would be interesting to give it a try to see if the problem still happens.

+tmathmeyer FYI