Texture mipmap data disapears after time or suspend.
Reported by
beto.c...@gmail.com,
Dec 5 2017
|
||||||||
Issue descriptionUserAgent: Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Ubuntu Chromium/62.0.3202.94 Chrome/62.0.3202.94 Safari/537.36 Steps to reproduce the problem: 1. Open https://twgljs.org/examples/2d-array-texture.html 2. suspend computer and wake up What is the expected behavior? nothing changes in example What went wrong? Farther from camera pixels where mipmaps are used are white or with random dots. Did this work before? N/A Does this work in other browsers? N/A Chrome version: 62.0.3202.94 Channel: stable OS Version: 62.0.3202.94 Flash Version: Same thing happens when you leave example in tab and return to it after some time (1h) using other tabs.
,
Dec 5 2017
,
Dec 8 2017
Reporter is using Linux/NVIDIA. I'm able to reproduce on my Linux/NVIDIA workstation. Unknown whether this is our bug or NVIDIA's.
,
Dec 8 2017
Since this didn't repro on Linux/Intel it seems likely this is Linux/NVIDIA specific. +oetuaho
,
Dec 8 2017
A possible workaround is to use the power monitor to lose webgl contexts on Linux/NVIDIA after a wake from suspend. FYI, I also saw similar corruption in my UE4 Editor window after suspend-cycling my machine.
,
Dec 8 2017
A little more clarification: the PowerObserver which is already in the GPU process could be used to watch for the suspend/resume events, and a new driver bug workaround could be added like "lose_contexts_upon_suspend_and_resume" which would be used only on Linux/NVIDIA. The difficulty here is that we would have to indicate that it's safe to restore this context. Currently all WebGL contexts stay lost if there was a "real" context loss coming from the GPU process. That bit of information would have to be passed back from the GPU process to the renderer process. https://cs.chromium.org/chromium/src/base/power_monitor/power_observer.h?q=PowerObserver&sq=package:chromium&l=13
,
Oct 9
,
Oct 18
I know this is an old issue, but I believe I have some more information that might help out with finding a solution to the issue. I am using a linux/Nvidia laptop which was able to repro the issue described. I repro'd the given repro in both Chrome and Firefox. I first saw this issue through VS Code. Google Chrome: 69.0.3497.100 (Official Build) (64-bit) Revision: 8920e690dd011895672947112477d10d5c8afb09-refs/branch-heads/3497@{#948} Chrome UA: Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/69.0.3497.100 Safari/537.36 Attached are relevant graphics information for both Chrome and FireFox.
,
Oct 19
#10: what laptop model exactly do you have and what Linux distribution and version are you running? Thanks. |
||||||||
►
Sign in to add a comment |
||||||||
Comment 1 by zmo@chromium.org
, Dec 5 2017