[SOLVED] screen tearing, mouse issues

chrisco23 · 2024-04-17 22:52:35

not sure which update of what might have broken everything but something in the last 24 hours.

sway screen tearing everywhere and almost impossible to type.

I understand that's vague but maybe somebody can tell me if there is a known April 16 breakage?

otherwise, maybe Sway? But Sway has worked flawlessly for ever since I migrated from i3.

this is a dell xps 15 btw.

thx

Last edited by chrisco23 (2024-04-26 16:53:03)

seth · 2024-04-18 08:54:52

this is a dell xps 15 btw.

"lspci -k" would be more helpful itr. btw.

What was updated in the last 24h?
(pacman log)
Sway? Do you get the same tearing in eg. weston? Or i3?

chrisco23 · 2024-04-18 15:00:09

thanks, and I haven't had a moment to look properly, have not tried switching over to i3, but fwiw here is recent pacman log...

[2024-04-17T17:25:24-0700] [PACMAN] Running 'pacman -R -n -s --config /etc/pacman.conf -- chromium'
[2024-04-17T17:25:27-0700] [ALPM] transaction started
[2024-04-17T17:25:27-0700] [ALPM] removed chromium (124.0.6367.60-1)
[2024-04-17T17:25:27-0700] [ALPM] transaction completed
[2024-04-17T17:25:27-0700] [ALPM] running '30-systemd-update.hook'...
[2024-04-17T17:25:27-0700] [ALPM] running 'gtk-update-icon-cache.hook'...
[2024-04-17T17:25:27-0700] [ALPM] running 'update-desktop-database.hook'...
[2024-04-17T17:25:38-0700] [PACMAN] Running 'pacman -R -n -s --config /etc/pacman.conf -- chromium'
[2024-04-17T17:25:45-0700] [PACMAN] Running 'pacman -S --config /etc/pacman.conf -- extra/chromium'
[2024-04-17T17:25:47-0700] [ALPM] transaction started
[2024-04-17T17:25:48-0700] [ALPM] installed chromium (124.0.6367.60-1)
[2024-04-17T17:25:48-0700] [ALPM] transaction completed
[2024-04-17T17:25:48-0700] [ALPM] running '30-systemd-update.hook'...
[2024-04-17T17:25:48-0700] [ALPM] running 'gtk-update-icon-cache.hook'...
[2024-04-17T17:25:48-0700] [ALPM] running 'update-desktop-database.hook'...
[2024-04-17T17:25:48-0700] [PACMAN] Running 'pacman -D -q --asexplicit --config /etc/pacman.conf -- chromium'
[2024-04-17T17:59:43-0700] [PACMAN] Running 'pacman -U --config /etc/pacman.conf -- /home/melinuxuser/.cache/yay/downgrade/downgrade-11.3.0-1-any.pkg.tar.zst'
[2024-04-17T17:59:46-0700] [ALPM] transaction started
[2024-04-17T17:59:46-0700] [ALPM] installed downgrade (11.3.0-1)
[2024-04-17T17:59:46-0700] [ALPM] transaction completed
[2024-04-17T17:59:46-0700] [ALPM] running '30-systemd-update.hook'...
[2024-04-17T17:59:46-0700] [PACMAN] Running 'pacman -D -q --asexplicit --config /etc/pacman.conf -- downgrade'
[2024-04-17T18:00:28-0700] [PACMAN] Running 'pacman -U https://archive.archlinux.org/packages/n/nvidia/nvidia-550.67-2-x86_64.pkg.tar.zst'
[2024-04-17T18:00:53-0700] [ALPM] transaction started
[2024-04-17T18:00:53-0700] [ALPM] downgraded nvidia (550.67-8 -> 550.67-2)
[2024-04-17T18:00:53-0700] [ALPM] transaction completed
[2024-04-17T18:00:53-0700] [ALPM] running '30-systemd-update.hook'...
[2024-04-17T18:00:53-0700] [ALPM] running '60-depmod.hook'...
[2024-04-17T18:09:46-0700] [PACMAN] Running 'pacman -S -y --config /etc/pacman.conf --'
[2024-04-17T18:09:46-0700] [PACMAN] synchronizing package lists
[2024-04-17T18:09:53-0700] [PACMAN] Running 'pacman -S -y -u --config /etc/pacman.conf --'
[2024-04-17T18:09:53-0700] [PACMAN] synchronizing package lists
[2024-04-17T18:09:54-0700] [PACMAN] starting full system upgrade
[2024-04-17T18:10:12-0700] [ALPM] transaction started
[2024-04-17T18:10:12-0700] [ALPM] upgraded electron29 (29.3.0-1 -> 29.3.1-1)
[2024-04-17T18:10:12-0700] [ALPM] upgraded electron30 (30.0.0-1 -> 30.0.1-1)
[2024-04-17T18:10:13-0700] [ALPM] upgraded nvidia (550.67-2 -> 550.67-8)
[2024-04-17T18:10:13-0700] [ALPM] transaction completed
[2024-04-17T18:10:13-0700] [ALPM] running '30-systemd-update.hook'...
[2024-04-17T18:10:13-0700] [ALPM] running '60-depmod.hook'...
[2024-04-17T18:10:15-0700] [ALPM] running 'update-desktop-database.hook'...
[2024-04-18T05:34:59-0700] [PACMAN] Running 'pacman -S -y --config /etc/pacman.conf --'
[2024-04-18T05:34:59-0700] [PACMAN] synchronizing package lists
[2024-04-18T05:35:06-0700] [PACMAN] Running 'pacman -S -y -u --config /etc/pacman.conf --'
[2024-04-18T05:35:06-0700] [PACMAN] synchronizing package lists
[2024-04-18T05:35:07-0700] [PACMAN] starting full system upgrade
[2024-04-18T05:35:12-0700] [ALPM] transaction started
[2024-04-18T05:35:12-0700] [ALPM] upgraded glibc (2.39-1 -> 2.39-2)
[2024-04-18T05:35:12-0700] [ALPM-SCRIPTLET] Generating locales...
[2024-04-18T05:35:13-0700] [ALPM-SCRIPTLET]   en_US.UTF-8... done
[2024-04-18T05:35:13-0700] [ALPM-SCRIPTLET] Generation complete.
[2024-04-18T05:35:13-0700] [ALPM] upgraded gmp (6.3.0-1 -> 6.3.0-2)
[2024-04-18T05:35:13-0700] [ALPM] upgraded lib32-glibc (2.39-1 -> 2.39-2)
[2024-04-18T05:35:13-0700] [ALPM] upgraded mpv (1:0.37.0-3 -> 1:0.38.0-2)
[2024-04-18T05:35:13-0700] [ALPM] transaction completed
[2024-04-18T05:35:13-0700] [ALPM] running '30-systemd-daemon-reload-system.hook'...
[2024-04-18T05:35:13-0700] [ALPM] running '30-systemd-update.hook'...
[2024-04-18T05:35:13-0700] [ALPM] running '80-cronie.hook'...
[2024-04-18T05:35:13-0700] [ALPM] running 'gtk-update-icon-cache.hook'...
[2024-04-18T05:35:13-0700] [ALPM] running 'texinfo-install.hook'...
[2024-04-18T05:35:14-0700] [ALPM] running 'update-desktop-database.hook'...

chrisco23 · 2024-04-18 15:01:03

may be irrelevant but all pacman things were resultant only from using "yay".

shoot sorry only now realizing I probably need to go further back. handicapped here.

Last edited by chrisco23 (2024-04-18 15:04:06)

seth · 2024-04-18 20:25:28

[2024-04-17T18:00:53-0700] [ALPM] downgraded nvidia (550.67-8 -> 550.67-2)

This doesn't work - the prebuilt OOT modules are bound to the specific kernel they were build against.
You'd also have to downgrade the kernel or use nvidia-dkms
You re-upped the module ~10m afterwards, but depending on your reboot sequence you might be running on the simpledrm or vesa driver right now.

chrisco23 · 2024-04-19 00:03:32

This morning I've kept things to my primary monitor (43" tv), and that seems *knock on wood* to allow me to at least type, but it's clearly not solved.

Chromium still ok at the moment it's mostly functioning, as long as I'm on this one screen, but even then, it's shaky, and I barely tested much before the screen goes haywire with the mouse and I just ... can't do much with it at this point.

Do you have some suggestion as to where to continue my troubleshooting?

For the moment I'm a little afraid of messing with stuff unless I understand better, because I could end up locked out of life. ...

thanks again

chrisco23 · 2024-04-19 00:04:44

the only thing I see when I run Chromium & that looks weird is this:

[2903:2930:0418/164724.774535:ERROR:shared_image_manager.cc(325)] SharedImageManager::ProduceMemory: Trying to Produce a Memory representation from a non-existent mailbox.

but I hardly believe that's helpful. It's at least not helping me much so far.

seth · 2024-04-20 16:29:52

Please post your complete system journal for the boot:

sudo journalctl -b | curl -F 'file=@-' 0x0.st

If you cannot access the log from the live system, see https://wiki.archlinux.org/title/System … al_to_view
But not "-e", but "-b -1"

chrisco23 · 2024-04-21 13:03:17

https://0x0.st/XogA.txt

seth · 2024-04-21 14:48:28

Apr 21 05:58:56 xps9520 kernel: [drm] Initialized simpledrm 1.0.0 20200625 for simple-framebuffer.0 on minor 0
Apr 21 05:58:56 xps9520 kernel: simple-framebuffer simple-framebuffer.0: [drm] fb0: simpledrmdrmfb frame buffer device

Enable https://wiki.archlinux.org/title/NVIDIA … de_setting - use the "nvidia_drm.modeset=1" kernel parameter (modprobe.conf won't do!)
But this actually isn't a VGA device, the relevant chip is the intel one

Apr 21 05:58:57 xps9520 kernel: i915 0000:00:02.0: [drm] *ERROR* [CONNECTOR:281:DP-4] Unknown HDMI VIC: 0
Apr 21 05:58:57 xps9520 kernel: fbcon: i915drmfb (fb0) is primary device
Apr 21 05:58:57 xps9520 kernel: snd_hda_intel 0000:00:1f.3: bound 0000:00:02.0 (ops i915_audio_component_bind_ops [i915])
Apr 21 05:58:57 xps9520 kernel: i915 0000:00:02.0: [drm] fb0: i915drmfb frame buffer device
Apr 21 05:58:58 xps9520 kernel: i915 0000:00:02.0: [drm] *ERROR* [CONNECTOR:281:DP-4] Unknown HDMI VIC: 0
Apr 21 05:58:58 xps9520 kernel: i915 0000:00:02.0: [drm] *ERROR* [CONNECTOR:281:DP-4] Unknown HDMI VIC: 0
Apr 21 05:59:15 xps9520 kernel: i915 0000:00:02.0: [drm] *ERROR* [CONNECTOR:281:DP-4] Unknown HDMI VIC: 0

but I guess the affected output is actually the notebook panel?

We'll just throw everything against the wall:

i915.enable_psr=0 i915.enable_fbc=0 intel_idle.max_cstate=1 i915.enable_dc=0 ahci.mobile_lpm_policy=1

https://wiki.archlinux.org/title/Kernel_parameters

If this works, you want to narrow down on the actually critical parameter - otherwise, curveball, disable the wifi.

chrisco23 · 2024-04-23 22:29:37

I had to step away for a time but I'm still struggling to understand the first step.

I've envountered this nvidia_drm.modeset=1 somewhere before.

I'm just not untderstanding what the docs are asking me to do when they say: "To enable it, set the modeset=1 kernel module parameter for the nvidia_drm module."

I also didn't really understand what you meant about the VGA device and the "notebook panel".

Everything has worked fine on my setup that I update daily since at least 6 months.

And I'm sorry I haven't gotten a handle on the bit about "i915.enable_psr=0 i915.enable_fbc=0 intel_idle.max_cstate=1 i915.enable_dc=0 ahci.mobile_lpm_policy=1
"

It's so hard because just trying to type it's like somebody is constantly holding down the backspace key.

One other thing occurred to me is that the only unusual thing I did recently was install Android Studio SDK and built a QEMU hackintosh. I don't know if that means anything at all, but I wonder if I could get rid of all that stuff. I've had to back off from trying to volunteer for mobile development.

Been sick so maybe I just can't grok all this until I feel better, so that is all I got so far.

seth · 2024-04-23 22:57:22

I also didn't really understand what you meant about the VGA device and the "notebook panel".

The nvidia GPU cannot be attached to a monitor and I assume you're perceiving the tearing on the notebook panel, not an external monitor?

It's so hard because just trying to type it's like somebody is constantly holding down the backspace key.

That symptom is absent from the OP?
Is it the same if you don't start sway but stay on a linux console? Do you get the tearing there as well? Can you ssh into the system?
Are there spurious event in "libinput debug-events" (ie. is there really a backspace event?)

I haven't gotten a handle on the bit about…

As indicated, those are https://wiki.archlinux.org/title/Kernel_parameters and you're supposed to add them.

chrisco23 · 2024-04-24 18:11:57

I see. Sorry I wasn't clear on some parts. I've been out of commission, offline.

The good news is things are looking much better if not maybe solved even. I want to test a little further before I come back and mark solved.

I used your kernel parameters. I just wasn't totally sure I did the right thing by inventing a filename (with .conf) to put in modprobe.d.

I don't see the backspace thing either this morning [knock]. That seemed like it would be something with my RK64 keyboard, or wireless mouse, so I was switching those in and out.

I intend to finally get up and at my computer some later today.

About panel, I hardly ever open the Dell XPS 15 machine at all. I use a cheap 43" 4K Tv as primary monitor (and where I saw the tearing). I use an old 24" Acer as secondary monitor.

I'll get back with more conclusive report by tonight, but so far today looking good. Caveat that I've only been in bed watching youtube videos (was sick a few days).

Thanks again.

seth · 2024-04-24 21:28:20

I just wasn't totally sure I did the right thing by inventing a filename (with .conf) to put in modprobe.d.

The kernel parameters would be illegal syntax in that context.

chrisco23 · 2024-04-25 02:39:18

I spoke too soon earlier. Problems still exist. But after another hour or two of looking more into things, trying some things, current status:

1. Firefox seems ok so far. I can live with that. Mainly used Chromium for the devtools (historical reasons) and because I use gmail and google calendar.
2. Qutebrowser also seems ok so far but I only tried briefly. It's too much to sacrifice extensions
3. Chrome almost works but there is the flicker / tearing sometimes, plus the keyboard randomly backspacing or double letters
4. Chromium was kinda working, at least starting when run from terminal (foot), but now that's somehow not working, black screen
5. Visual Studio Code exhibits the keyboard issue but no tearing, understandable for VSC though

The kernel stuff I had tried to do when my head wasn't too clear (was sick) was:
cat nvidia_fix_april_22.conf
options nvidia_drm.modeset=1 i915.enable_psr=0 i915.enable_fbc=0 intel_idle.max_cstate=1 i915.enable_dc=0 ahci.mobile_lpm_policy=1

I have since moved that out of modprobe.d, set it aside since I'm guessing I prob. did something wrong there.

I removed some electron25 stuff that I believe got installed when I was messing with docker-osx and the hackintosh stuff.
Don't understand why I need half dozen versions of electron but not worth digging into all that dependency hell.
For same reason, I removed nvidia-container-utils or whatever.

These last two removals I don't see any difference though.

So I believe I can get by with Firefox for now.
I can get by without VSC, but not too long hopefully.

So I'll keep this open yet and see if maybe an update or something comes along to help Chromium and Chrome in the next day or so.

I've already taken up a lot of your time here. Thanks again. If you happen to think of anything else given the above info, I'm up and at least I'm functioning better myself.

seth · 2024-04-25 09:10:34

The kernel stuff I had tried to do when my head wasn't too clear (was sick) was:
cat nvidia_fix_april_22.conf
options nvidia_drm.modeset=1 i915.enable_psr=0 i915.enable_fbc=0 intel_idle.max_cstate=1 i915.enable_dc=0 ahci.mobile_lpm_policy=1

I have since moved that out of modprobe.d, set it aside since I'm guessing I prob. did something wrong there.

Yes. Actually add those to the kernel command line and see whether that has an effect.
If you did it right, those tokens will show up in "cat /proc/cmdline"?

chrisco23 · 2024-04-26 16:57:06

Variously trying things I've had things all working for 24-hours now, so marking solved.

The thing that appeared to have solved me was switching from nvidia to nvidia-dkms.

I'd still like to know why I needed to do that, if I should do that, etc. I looked a little to see if I could find any module or something that would necessitate the dkms part, but I didn't see anything. I may have been on nvidia-dkms before and got switched to nvidia at some point during my aborted attempt to set up a mobile dev environment here.

Hopefully the next update doesn't cause something to happen between nvidia update and kernel update.

Nvidia: Never again. The only reason I have Nvidia still is because 1. I'd been using an earlier Dell XPS 15 for about 6-7 years; and 2. I was in a hurry 6 months ago when I bought this newer one.

seth · 2024-04-26 19:58:10

The thing that appeared to have solved me was switching from nvidia to nvidia-dkms.

This is rather unlikely, but nvidia saw a minor version updated on Apr 19 (550.67 => 550.76)

chrisco23 · 2024-04-26 20:30:00

Oh no. I just had to let curiosity get the best of me.

I did "yay -Rns nvidia-dkms" (then allowed replacement)
and "Yay -s nvidia"

That brought me back to the problems of yesterday, so I confidently did "Yay -S nvidia-dkms" (then allowed replacement)

Now I'm still back to the original problems.

Unbelievable... don't fix it if it ain't broke, for sure

seth · 2024-04-26 20:34:31

Don't use yay…

I've had things all working for 24-hours now

How often did you reboot during that period?

Please post journal from both boots (good nvidia-dkms and bad nvidia-dkms), eg.

sudo journalctl -b | curl -F 'file=@-' 0x0.st # current one
sudo journalctl -b -2 | curl -F 'file=@-' 0x0.st # current one  # 2 boots ago, you might have to go one or two boots back in time more

chrisco23 · 2024-04-26 20:39:14

Well I don't know how to get any "good nvidia-dkms" now. But here are the results:

http://0x0.st/XH9u.txt

http://0x0.st/XH9S.txt

chrisco23 · 2024-04-26 20:48:47

ok, I *think* I'm back in business. I think my problem had been compounded because I use fish as a daily driver and I had an alias for chromium (chromium --ozone-platform=wayland &). That alias/function has worked fine for more months than I can remember.

I found the proper way to unalias. I know, you will probably say "don't use fish". As for "don't use yay", why would that be a problem? I've used yay for years.

seth · 2024-04-26 21:20:18

I'll rather say "what"?
So you had configured chromium to explicitly use the native wayland backend, which because of a recent regression would actually have been required to make it wayland at all.
So where would have been the problem with this? Silently lost the alias co-incidental w/ your nvidia driver re-installation and ran chromium on xwayland?

Wrt yay:
https://wiki.archlinux.org/title/AUR_he … n_wrappers
In the specific case of yay, this might just be typos in your post, but https://bbs.archlinux.org/viewtopic.php … 0#p2148350

chrisco23 · 2024-04-26 21:51:43

In the past, I've been warned/chastised about using fish, which I understand makes sense when troubleshooting. When I need to, or think I need to, then I run bash. I recently installed zsh but not used it really because I don't need any more customization at this point. Fish shell is just too convenient.

As for the the chromium alias (fish calls these "functions"), I really don't know why the alias would have been problematic as it always used to work, but at this point I don't think I need any alias/function there anymore.

The Chromium problem I was having was a problem whether I tried to launch from wofi (my usual way), or command line.

In any case, I was having a lot more than just the Chromium problem (the keyboard thing for one, also seems fixed).

More than once I've spoken too soon saying things are resolved, but hopefully nothing changes now, nothing moves...

seth · 2024-04-26 22:01:06

https://wiki.archlinux.org/title/Waylan … s_visually

I'll speculate that the behavor is simply non-deterministic, the actual cause not identified and changes reg. nvidia/dkms and some shell alias being merely coincidental.

Arch Linux

#1 2024-04-17 22:52:35

[SOLVED] screen tearing, mouse issues

#2 2024-04-18 08:54:52

Re: [SOLVED] screen tearing, mouse issues

#3 2024-04-18 15:00:09

Re: [SOLVED] screen tearing, mouse issues

#4 2024-04-18 15:01:03

Re: [SOLVED] screen tearing, mouse issues

#5 2024-04-18 20:25:28

Re: [SOLVED] screen tearing, mouse issues

#6 2024-04-19 00:03:32

Re: [SOLVED] screen tearing, mouse issues

#7 2024-04-19 00:04:44

Re: [SOLVED] screen tearing, mouse issues

#8 2024-04-20 16:29:52

Re: [SOLVED] screen tearing, mouse issues

#9 2024-04-21 13:03:17

Re: [SOLVED] screen tearing, mouse issues

#10 2024-04-21 14:48:28

Re: [SOLVED] screen tearing, mouse issues

#11 2024-04-23 22:29:37

Re: [SOLVED] screen tearing, mouse issues

#12 2024-04-23 22:57:22

Re: [SOLVED] screen tearing, mouse issues

#13 2024-04-24 18:11:57

Re: [SOLVED] screen tearing, mouse issues

#14 2024-04-24 21:28:20

Re: [SOLVED] screen tearing, mouse issues

#15 2024-04-25 02:39:18

Re: [SOLVED] screen tearing, mouse issues

#16 2024-04-25 09:10:34

Re: [SOLVED] screen tearing, mouse issues

#17 2024-04-26 16:57:06

Re: [SOLVED] screen tearing, mouse issues

#18 2024-04-26 19:58:10

Re: [SOLVED] screen tearing, mouse issues

#19 2024-04-26 20:30:00

Re: [SOLVED] screen tearing, mouse issues

#20 2024-04-26 20:34:31

Re: [SOLVED] screen tearing, mouse issues

#21 2024-04-26 20:39:14

Re: [SOLVED] screen tearing, mouse issues

#22 2024-04-26 20:48:47

Re: [SOLVED] screen tearing, mouse issues

#23 2024-04-26 21:20:18

Re: [SOLVED] screen tearing, mouse issues

#24 2024-04-26 21:51:43

Re: [SOLVED] screen tearing, mouse issues

#25 2024-04-26 22:01:06

Re: [SOLVED] screen tearing, mouse issues

Board footer