UI (entire) becomes unresponsive

Description of the issue:
Debian Linux 10 on Thinkpad w/ integrated Intel graphics
Brave repo entry: deb [arch=amd64] https://brave-browser-apt-release.s3.brave.com/ buster main
using Xorg + window manager (jwm)

For the last few days (perhaps as long as a week). Brave will occasionally become completely unresponsive—cannot scroll or interact with buttons/menus, even keyboard shortcuts are ignored. Indeed, the mouse cursor no longer changes when moving over the address bar and the like. The rest of my system continues to behave normally save for the parts of my screen where dead brave windows cannot be moved or hidden. Brave must be forcibly quit (pkill) and restarted to be usable again. This has happened more than a dozen times. It has happened while I’ve been actively using Brave and I’ve also returned to my computer to find Brave already frozen in this way (having been normal when I last used it). I’ve seen one update go through while this issue has been happening, so if it was an update that precipitated the issue it will have been at least 2 updates ago.
How can this issue be reproduced?

  1. Open Brave browser
  2. use it for awhile (or just leave it sitting open)

Expected result:
Application remains usable
Brave Version( check About Brave):
Version 1.38.115 Chromium: 101.0.4951.64 (Official Build) (64-bit)
Additional Information:
First noticed this within the last week. I update manually via apt fairly regularly (every couple of days) so I can’t pinpoint the exact update when this issue began, but it was very recent.

@JReis ,

Try . . . Disable hardware acceleration.

In a Brave Browser > New Window, go to: brave://settings/system

Will test and report back if the issue recurs or if I make it through tomorrow’s work-day without a problem.

Well, that does indeed seemed to have rectified the issue. I got through a full work day without a freeze. My thanks for the suggestion.

Should I mark this workaround as a solution? Obviously hardware acceleration had been working on this machine for months if not years. Feels like a regression, certainly.

1 Like

@JReis ,

Glad that worked for you.

Not a regression. Instead, there is an optimal performance range - of rates for a GPU processing instructions - while maintaining some cool, as in, not causing problems.

When a GPU is ‘thirsty for more’ and fed more, it gets hot, and a variety of symptoms show. When the screen flickers, that sometimes means, that a limiter has stepped in, to throttle the GPU and maintain some cool.

Yet, that limiter may not be enough, and the user has to step in and Disable Hardware Acceleration.

Something like that.