"Firefox" browsers tabs go unresponsive

jamess7995 · 2023-03-28 19:33:24

I use Firefox, Firedragon, and LibreWolf. All 3 exhibit the same behavior. I have 32GB of ram, and I use my desktop for personal home lab dev work, personal stuff like banking, schoolwork, and gaming. I like to have the separate browsers to compartmentalize things. I don't use librewolf for anything but schoolwork. If I'm in the middle of dev work and want to switch to schoolwork, I just open libre wolf, leaving the other browser open. All three of them inconsistently, for 2 years, even after an unrelated reinstall, will go unresponsive, kinda. Tab 1 could be puppies, tab 2 - kittens, and tab 3 - goldfish. I'll try to switch from tab 1 to tab 2. The window below the bookmarks bar continues displaying tab 1 stuff, the puppies. The title bar, level with the close window button, will change to kittens however. Sometimes it will eventually switch the displayed content, if I click enough, not if I wait long enough. The only resolution is to mess with the window. This might mean closing it and relaunching. Using CTRL + N, taking the tab out of that window, now it's separated in its own window, and apparently adding it back still fixes it. Sometimes I go weeks without experiencing this. Sometimes it happens a few times a day. Sometimes it happens when I've opened a total of 3 tabs, sometimes it happens when I open 30 tabs. This isn't necessarily back to back, sometimes it is. I just discovered that moving the tab out and back in "resolves" it. I was hoping to seek advice on fixing it while it was occurring, but now it might be gone for weeks or 2 minutes. What path do I take to resolve this kind of thing. I've tried to find posts about this, surely I'm not the only one with this problem, maybe I'm describing it weirdly.

arch_jsb · 2023-03-28 19:35:48

following

seth · 2023-03-28 21:00:57

@arch_jsb, there's a scubscribe link in the lower right corner of every thread.

@jamess7995 … and an Enter key on your keyboard.
To isolate

the relevant part wrote:

Tab 1 could be puppies, tab 2 - kittens, and tab 3 - goldfish. I'll try to switch from tab 1 to tab 2.
The window below the bookmarks bar continues displaying tab 1 stuff, the puppies. The title bar, level with the close window button, will change to kittens however.
Sometimes it will eventually switch the displayed content, if I click enough, not if I wait long enough.
The only resolution is to mess with the window.

Does the troubleshoot mode exhibit this behavior?

killall firefox; sleep 5; firefox -safe-mode

What display server is this (X11 or some wayland compositor)?
In case of X11, please post your Xorg log, https://wiki.archlinux.org/title/Xorg#General
And on a limb, remove xf86-video-intel,

grep -ri intel /{etc,usr/share}/X11/xorg.conf*

to make sure it's also not referenced in any config file (or Xorg won't start)

Last edited by seth (2023-03-28 21:01:21)

jamess7995 · 2023-03-28 21:36:08

seth wrote:

Does the troubleshoot mode exhibit this behavior?
killall firefox; sleep 5; firefox -safe-mode

I will have to start using safe mode to test. It will be tricky since I sometime go weeks without experiencing the issue. That's not to say I won't, but I might not find out quickly.

In case of X11, please post your Xorg log, https://wiki.archlinux.org/title/Xorg#General

I'm using X11 and KDE.
I pasted using ix.io. I didn't know if it would be obnoxious to paste the whole Xorg file. I've provided two versions of the file, the first is the latest 1000 lines, tail -n 1000. The second version is the full file.
shorter version
longer_version
If the log needs to be checked closer to the issue occurring, I'll now know to check that next time it occurs.

And on a limb, remove xf86-video-intel,
grep -ri intel /{etc,usr/share}/X11/xorg.conf*
to make sure it's also not referenced in any config file (or Xorg won't start)

I went to remove xf86-video-intel, it's not found, which makes sense. I'm using nvidia drivers.
I used that grep command and received 0 output, which makes sense to me since that shouldn't be installed.

Next time it occurs, hopefully I'll be in safemode, and I'll check and report back with terminal info from launching safe mode, as well as an updated xorg log. I hope I've provided what you wanted for now.

seth · 2023-03-28 22:14:45

nvidia GL contexts have a tendency to freeze when you switch the framebuffer control, typically after waking from S3 (suspend to ram) or switching to a different console (ctrl+alt+f2,3,4…) and back.

jamess7995 · 2023-03-28 22:22:56

seth wrote:

nvidia GL contexts have a tendency to freeze when you switch the framebuffer control, typically after waking from S3 (suspend to ram) or switching to a different console (ctrl+alt+f2,3,4…) and back.

I think we can rule that out if I'm understanding what you're saying. I don't put my desktop to sleep. I do have kvm switches to use a work provided imac for work from home. I enter those tty consoles once or twice a year when I mess my system up. The issue also occurs while I'm actively using the browser. It hasn't been the kind of thing where I've walked away and come back to issues. It's always been I'm in the middle of doing something, I go to switch tabs, and the window doesn't update properly.

seth · 2023-03-29 07:23:57

Are we in reality talking about 3 tabs, 30, 300 or 3000?
Does resizing the window immediately "fix" it or does it still take a moment?

jamess7995 · 2023-03-29 18:42:32

seth wrote:

Are we in reality talking about 3 tabs, 30, 300 or 3000?
Does resizing the window immediately "fix" it or does it still take a moment?

Not 3 tabs. I've had it happen with a small amount like 7 tabs, I've had it happen with 30 tabs. It doesn't appear tied to the amount of tabs, I'll sometimes have a bunch, as in like 20 tabs, per browser, and it'll be fine. Sometimes I have only one browser open, only a few tabs, around 7, and the issue can occur. I wanna say that resizing the window has been tried but I don't know with 100% certainty, I'll check that next time.

jamess7995 · 2023-04-12 18:53:12

seth wrote:

Are we in reality talking about 3 tabs, 30, 300 or 3000?
Does resizing the window immediately "fix" it or does it still take a moment?

Resizing has 0 impact. I was able to reproduce it in librewolf. In safe mode. This is the output, which seems irrelevant:

[Parent 39534, Main Thread] WARNING: dbusmenu_menuitem_property_set_shortcut: assertion 'gtk_accelerator_valid(key, modifier)' failed: 'glib warning', file /builds/librewolf-community/browser/arch/src/firefox-111.0.1/toolkit/xre/nsSigHandlers.cpp:167

(LibreWolf:39534): LIBDBUSMENU-GTK-CRITICAL **: 18:03:56.405: dbusmenu_menuitem_property_set_shortcut: assertion 'gtk_accelerator_valid(key, modifier)' failed
[Parent 39534, Main Thread] WARNING: dbusmenu_menuitem_property_set_shortcut: assertion 'gtk_accelerator_valid(key, modifier)' failed: 'glib warning', file /builds/librewolf-community/browser/arch/src/firefox-111.0.1/toolkit/xre/nsSigHandlers.cpp:167

(LibreWolf:39534): LIBDBUSMENU-GTK-CRITICAL **: 18:05:08.030: dbusmenu_menuitem_property_set_shortcut: assertion 'gtk_accelerator_valid(key, modifier)' failed
[Parent 39534, Main Thread] WARNING: dbusmenu_menuitem_property_set_shortcut: assertion 'gtk_accelerator_valid(key, modifier)' failed: 'glib warning', file /builds/librewolf-community/browser/arch/src/firefox-111.0.1/toolkit/xre/nsSigHandlers.cpp:167

(LibreWolf:39534): LIBDBUSMENU-GTK-CRITICAL **: 18:51:57.666: dbusmenu_menuitem_property_set_shortcut: assertion 'gtk_accelerator_valid(key, modifier)' failed

Here's the xorg log

seth · 2023-04-12 20:40:43

[    28.819] (**) Option "NoPM" "true"

??

Generally the xorg.conf is pointless cruft.

nvidia-settings -q all

(That's gonna be a shit-ton of text, so better redirect it into a file or pastebin service)

jamess7995 · 2023-04-12 21:35:53

seth wrote:

[    28.819] (**) Option "NoPM" "true"
??
Generally the xorg.conf is pointless cruft.
nvidia-settings -q all
(That's gonna be a shit-ton of text, so better redirect it into a file or pastebin service)

I've no clue what NoPM means, here's the nvidia settings

seth · 2023-04-13 06:02:44

https://man.archlinux.org/man/extra/xor … #Option~14
Has this ever happened w/ only one monitor (preferably not the 165Hz one) attached?

jamess7995 · 2023-04-13 14:42:18

seth wrote:

https://man.archlinux.org/man/extra/xor … #Option~14
Has this ever happened w/ only one monitor (preferably not the 165Hz one) attached?

I've always had at least 2 monitors connected. If this issue occurred daily, I could totally just disconnect 2/3 monitors but with how rare this could be, I'd rather just let it go than disconnect the monitors. I'm open to other ideas for investigation.

seth · 2023-04-13 14:55:07

Disconnect at least the 165Hz one or run it at 60Hz.

jamess7995 · 2023-04-13 15:08:53

seth wrote:

Disconnect at least the 165Hz one or run it at 60Hz.

I would be much more willing to do that, I can tell you though, this issue occurred before I added monitor #3, the 165HZ one. I'll still disconnect it while investigating.

seth · 2023-04-13 19:10:11

this issue occurred before I added monitor #3

renders

I'll still disconnect it

pointless.

If this issue occurred daily … but with how rare this could be

So how frequent (roughly) is this? Can you make out any pattern that would explain why the GL context would have to be recreated (otherwise use of VRAM, de/activation of outputs, screensaver/DPMS, …)?

jamess7995 · 2023-04-14 18:31:34

seth wrote:

this issue occurred before I added monitor #3
renders
I'll still disconnect it
pointless.
If this issue occurred daily … but with how rare this could be
So how frequent (roughly) is this? Can you make out any pattern that would explain why the GL context would have to be recreated (otherwise use of VRAM, de/activation of outputs, screensaver/DPMS, …)?

Sometimes it occurs multiple times daily, sometimes it doesn't occur for a couple weeks. I use these browsers almost daily for 1-2 hours, but multiple times a week I'm actively using them for 6+ hours. I frequently will be searching a tab for some string, and will switch from tab to tab, using CTRL + F, and it will be fine, sure sometimes it does the behavior, but I've had it happen where I haven't touched the browser for 5 minutes, and I switch from one tab to another, and the issue presents itself. I'm sure there is a pattern, there's gotta be, but I haven't found it, its always felt inconsistent and random.

seth · 2023-04-14 19:47:31

Have you considered that the driving pattern might be external to the browser (eg. the forementioned S3, but also other VRAM intense clients might be relevant, daytime - eg. some cronjob…)

jamess7995 · 2023-04-15 23:54:45

seth wrote:

Have you considered that the driving pattern might be external to the browser (eg. the forementioned S3, but also other VRAM intense clients might be relevant, daytime - eg. some cronjob…)

I don't really know what to consider, I'm not sure how to diagnose or troubleshoot this. I know it occurs in different firefox browsers, I don't use chromium unless I have to for some specific website I encounter. I know I haven't had this issue in other programs. I wouldn't know where to go from here, that's why I'm here.

seth · 2023-04-16 07:13:28

The symptoms sound like what's common w/ the nvidia blob when changing the framebuffer control, typically on resume from suspend or simply changing to a different VT.
We're looking for anything else that might encounter at the time and maybe relate to the framebuffer control.

To make sure it's the HW and not some freakish bug in FF:
1. disable all HW acceleration in FF and see whether the issue shows up again w/o
2. only disable the webrenderer HW accel and see whether the issue shows up again w/o
3. inspect the dmesg/journal fro nvidia/nvrm driver messages (notably GPU resets, they're not common to the framebuffer control issue, though.

Also make sure you're setting "nvidia-drm.modeset=1", https://wiki.archlinux.org/title/NVIDIA … de_setting

Arch Linux

#1 2023-03-28 19:33:24

"Firefox" browsers tabs go unresponsive

#2 2023-03-28 19:35:48

Re: "Firefox" browsers tabs go unresponsive

#3 2023-03-28 21:00:57

Re: "Firefox" browsers tabs go unresponsive

#4 2023-03-28 21:36:08

Re: "Firefox" browsers tabs go unresponsive

#5 2023-03-28 22:14:45

Re: "Firefox" browsers tabs go unresponsive

#6 2023-03-28 22:22:56

Re: "Firefox" browsers tabs go unresponsive

#7 2023-03-29 07:23:57

Re: "Firefox" browsers tabs go unresponsive

#8 2023-03-29 18:42:32

Re: "Firefox" browsers tabs go unresponsive

#9 2023-04-12 18:53:12

Re: "Firefox" browsers tabs go unresponsive

#10 2023-04-12 20:40:43

Re: "Firefox" browsers tabs go unresponsive

#11 2023-04-12 21:35:53

Re: "Firefox" browsers tabs go unresponsive

#12 2023-04-13 06:02:44

Re: "Firefox" browsers tabs go unresponsive

#13 2023-04-13 14:42:18

Re: "Firefox" browsers tabs go unresponsive

#14 2023-04-13 14:55:07

Re: "Firefox" browsers tabs go unresponsive

#15 2023-04-13 15:08:53

Re: "Firefox" browsers tabs go unresponsive

#16 2023-04-13 19:10:11

Re: "Firefox" browsers tabs go unresponsive

#17 2023-04-14 18:31:34

Re: "Firefox" browsers tabs go unresponsive

#18 2023-04-14 19:47:31

Re: "Firefox" browsers tabs go unresponsive

#19 2023-04-15 23:54:45

Re: "Firefox" browsers tabs go unresponsive

#20 2023-04-16 07:13:28

Re: "Firefox" browsers tabs go unresponsive

Board footer