When i use firefox, after a while, any key typing (not mouse moving or clicking) gets delayed by 20 seconds or more. No other running program is affected meanwhile (e.g. thunderbird, shell window). The issue can only be fixed by restarting firefox Reproducible: Always Steps to Reproduce: 1. start firefox 2. use it for some time, usually longer than 10 minutes Actual Results: For typing, firefox is hardly usable anymore e.g. while reporting a bug in this bugzilla Expected Results: firefox remains usable firefox-114.0.2-1.fc38.x86_64
Additional info: firefox' CPU utilization looks normal while the problem is visible. So it should not be busy i guess. Pasting text with the middle mouse button works normally as well. Just typing does not.
Can you please run on terminal with MOZ_LOG="Widget:5" env variable like: MOZ_LOG="Widget:5" firefox 2>&1 | grep Key and check if key events are logged in time or are delayed? Thanks.
Cannot reproduce it currently. I have wrapped the firefox program, so i always get key events logged, when firefox is started. Please feel free to close this issue for now. When it comes back, i'll reopen.
Okay, Thanks.
Recently i had seen that again, but due to whatever changes in updates of Fedora my wrapper was not active anymore, so i don't have any captured typing. Instead, i have observed this absurd phenomenon on thunderbird as well. Assuming, the code base is to some degree common, i have wrapped thunderbird the same way. Can anyone confirm, that the environment setting MOZ_LOG="Widget:5" is effective with thunderbird, too ?
(In reply to Albert Flügel from comment #5) > Recently i had seen that again, but due to whatever changes in updates of > Fedora my wrapper was not active anymore, so i don't have any captured > typing. > > Instead, i have observed this absurd phenomenon on thunderbird as well. > Assuming, the code base is to some degree common, i have wrapped thunderbird > the same way. Can anyone confirm, that the environment setting > MOZ_LOG="Widget:5" is effective with thunderbird, too ? Yes, it's the same codebase. Please check if the delay is on TB side. Thanks.