You are not logged in.

#26 2020-10-21 10:33:55

eblau
Member
Registered: 2016-02-11
Posts: 43

Re: [FIXED?] Chromium artefacts after resume from suspend or tty switch

@salkay, linux 5.9.1.arch1-1 is working for you? I saw the Arch news update that said not to upgrade to 5.9.x due to NVIDIA driver incompatibility:

https://forums.developer.nvidia.com/t/n … 5-9/157263

Offline

#27 2020-10-21 10:38:24

Salkay
Member
Registered: 2014-05-22
Posts: 618

Re: [FIXED?] Chromium artefacts after resume from suspend or tty switch

Arch news has more details. Nvidia graphics work fine overall, it's just CUDA and OpenCL that are broken, neither of which I use.

Offline

#28 2020-11-09 05:18:37

Sha256NaCl
Member
Registered: 2020-06-15
Posts: 1

Re: [FIXED?] Chromium artefacts after resume from suspend or tty switch

I have the same problem for weeks and I can't find a fix for it.
Vs-code is also effected by the same bug/glitch.

kernel: 5.9.6-arch1-1
gpu: nvidia quadro p1000
nvidia driver version: 455.38

Last edited by Sha256NaCl (2020-11-09 05:19:54)

Offline

#29 2020-11-19 16:49:46

hirnschmalz
Member
Registered: 2020-11-19
Posts: 36

Re: [FIXED?] Chromium artefacts after resume from suspend or tty switch

For me it helped to enable Vulkan for Chromium in chrome://flags/#enable-vulkan

Last edited by hirnschmalz (2020-11-19 16:50:00)

Offline

#30 2020-11-22 04:48:59

cubethethird
Member
Registered: 2016-01-25
Posts: 61
Website

Re: [FIXED?] Chromium artefacts after resume from suspend or tty switch

Just want to add (mainly for searchability of this issue) that this also seems to affect many electron apps, such as Code (VS Code), Discord, Caprine, Slack, etc.

Offline

#31 2020-11-22 07:06:30

kokoko3k
Member
Registered: 2008-11-14
Posts: 2,390

Re: [FIXED?] Chromium artefacts after resume from suspend or tty switch

Guys, apart from the experimental services already mentioned by V1del and enabling vulkan in chromium (i can't say why it is a workaround for the issue), there's really not much you can do apart from NOT using nvidia.
There are bug reports everywhere.
Some have implemented specific workarounds for nvidia drivers (plasma,kde).
The issue is about video memory corruption, not specific to chromium engine and yes, it affects suspend to ram, suspend to disk AND tty switch (and so multiple Xorg sessions too).
For the latter, experimental services wouldn't help anyway.

Last edited by kokoko3k (2020-11-22 07:16:27)


Help me to improve ssh-rdp !
Retroarch User? Try my koko-aio shader !

Offline

#32 2020-11-22 11:06:05

sarlej
Member
Registered: 2020-11-22
Posts: 1

Re: [FIXED?] Chromium artefacts after resume from suspend or tty switch

Here is one dirty workaround: link to askubuntu. But it works well.
If you kill browser gpu-process, browser will restart that process and artefacts are gone.
You can list process with:

#change vivaldi-bin to browser you like
pgrep -fa 'vivaldi-bin \-\-type=gpu-process'

You can run it automatically after resume. Make file /lib/systemd/system-sleep/revive-chrome-gpu

#!/bin/sh

set -e

if [ "$2" = "suspend" ] || [ "$2" = "hybrid-sleep" ]
then
    case "$1" in
        pre)
            true 
            ;;
        post) 
            sleep 1
            pkill -f 'vivaldi-bin \-\-type=gpu-process'
            ;;
    esac
fi

Change vivaldi-bin to browser you like, and make it executable.

Last edited by sarlej (2020-11-22 14:56:23)

Offline

#33 2020-11-23 11:39:10

eblau
Member
Registered: 2016-02-11
Posts: 43

Re: [FIXED?] Chromium artefacts after resume from suspend or tty switch

After updating two days back, this problem has gone away for me with chromium. I've done several suspend/resume cycles with no artifacts seen.

I'm not sure which package upgrade did the trick, but there are a lot of plausible candidates including the kernel, X.org, libdrm, chromium, etc.:

[2020-11-21T07:46:42-0500] [ALPM] upgraded acpica (20200925-1 -> 20201113-1)
[2020-11-21T07:46:42-0500] [ALPM] upgraded adobe-source-code-pro-fonts (2.030ro+1.050it-6 -> 2.032ro+1.052it+1.012var-1)
[2020-11-21T07:46:42-0500] [ALPM] upgraded alsa-card-profiles (13.99.3-1 -> 13.99.3+4+gd83ad6990-1)
[2020-11-21T07:46:42-0500] [ALPM] upgraded util-linux-libs (2.36-4 -> 2.36.1-3)
[2020-11-21T07:46:42-0500] [ALPM] upgraded asciidoc (9.0.3-1 -> 9.0.4-1)
[2020-11-21T07:46:42-0500] [ALPM] upgraded pam (1.4.0-3 -> 1.5.0-1)
[2020-11-21T07:46:42-0500] [ALPM] upgraded libcap (2.44-1 -> 2.45-2)
[2020-11-21T07:46:42-0500] [ALPM] upgraded ca-certificates-mozilla (3.58-2 -> 3.59-1)
[2020-11-21T07:46:42-0500] [ALPM] upgraded bbswitch-dkms (0.8-363 -> 0.8-364)
[2020-11-21T07:46:42-0500] [ALPM] upgraded c-ares (1.16.1-2 -> 1.17.1-1)
[2020-11-21T07:46:42-0500] [ALPM] upgraded nss (3.58-2 -> 3.59-1)
[2020-11-21T07:46:42-0500] [ALPM] upgraded libcups (2.3.3-3 -> 2.3.3+106+ga72b0140e-1)
[2020-11-21T07:46:42-0500] [ALPM] upgraded util-linux (2.36-4 -> 2.36.1-3)
[2020-11-21T07:46:42-0500] [ALPM] upgraded libpulse (13.99.3-1 -> 13.99.3+4+gd83ad6990-1)
[2020-11-21T07:46:42-0500] [ALPM] upgraded libdrm (2.4.103-1 -> 2.4.103-2)
[2020-11-21T07:46:42-0500] [ALPM] upgraded xorg-server-common (1.20.9-2 -> 1.20.9.r21.g5c400cae1-2)
[2020-11-21T07:46:42-0500] [ALPM] upgraded xorg-server (1.20.9-2 -> 1.20.9.r21.g5c400cae1-2)
[2020-11-21T07:46:43-0500] [ALPM] upgraded libibus (1.5.22+8+gf591381e-1 -> 1.5.23+1+gdd4cc5b0-1)
[2020-11-21T07:46:43-0500] [ALPM] upgraded librsvg (2:2.50.1-1 -> 2:2.50.2-1)
[2020-11-21T07:46:43-0500] [ALPM] upgraded js78 (78.4.1-1 -> 78.5.0-1)
[2020-11-21T07:46:43-0500] [ALPM] upgraded xorg-xprop (1.2.4-2 -> 1.2.5-1)
[2020-11-21T07:46:43-0500] [ALPM] upgraded chromium (86.0.4240.198-1 -> 87.0.4280.66-1)
[2020-11-21T07:46:44-0500] [ALPM] upgraded cmake (3.18.4-1 -> 3.19.0-1)
[2020-11-21T07:46:44-0500] [ALPM] upgraded imagemagick (7.0.10.37-1 -> 7.0.10.40-1)
[2020-11-21T07:46:44-0500] [ALPM] upgraded cups-filters (1.28.5-1 -> 1.28.5-3)
[2020-11-21T07:46:44-0500] [ALPM] upgraded cups (2.3.3-3 -> 2.3.3+106+ga72b0140e-1)
[2020-11-21T07:46:45-0500] [ALPM] upgraded evolution-data-server (3.38.1-1 -> 3.38.2-1)
[2020-11-21T07:46:45-0500] [ALPM] upgraded gpgme (1.14.0-1 -> 1.15.0-1)
[2020-11-21T07:46:46-0500] [ALPM] upgraded evolution (3.38.1-1 -> 3.38.2-1)
[2020-11-21T07:46:46-0500] [ALPM] upgraded evolution-ews (3.38.1-1 -> 3.38.2-1)
[2020-11-21T07:46:46-0500] [ALPM] upgraded firefox (82.0.3-1 -> 83.0-1)
[2020-11-21T07:46:46-0500] [ALPM] upgraded geoip-database (20200929-1 -> 20201110-1)
[2020-11-21T07:46:46-0500] [ALPM] upgraded gtk-doc (1.33.0-1 -> 1.33.1-1)
[2020-11-21T07:46:46-0500] [ALPM] upgraded vim-runtime (8.2.1704-1 -> 8.2.1989-2)
[2020-11-21T07:46:46-0500] [ALPM] upgraded gvim (8.2.1704-1 -> 8.2.1989-2)
[2020-11-21T07:46:47-0500] [ALPM] upgraded highlight (3.58-1 -> 3.59-1)
[2020-11-21T07:46:47-0500] [ALPM] upgraded intel-gmmlib (20.3.2-1 -> 20.3.3-1)
[2020-11-21T07:46:47-0500] [ALPM] upgraded intel-media-driver (20.3.0-1 -> 20.4.1-1)
[2020-11-21T07:46:47-0500] [ALPM] upgraded intel-ucode (20201112-1 -> 20201118-1)
[2020-11-21T07:46:47-0500] [ALPM] upgraded lib32-util-linux (2.36-1 -> 2.36.1-1)
[2020-11-21T07:46:47-0500] [ALPM] installed lib32-libtirpc (1.2.6-1)
[2020-11-21T07:46:47-0500] [ALPM] installed lib32-libnsl (1.3.0-1)
[2020-11-21T07:46:47-0500] [ALPM] installed lib32-pam (1.4.0-1)
[2020-11-21T07:46:47-0500] [ALPM] upgraded lib32-libcap (2.44-1 -> 2.45-1)
[2020-11-21T07:46:47-0500] [ALPM] upgraded libfbclient (3.0.4.33054-2 -> 3.0.6.33328-1)
[2020-11-21T07:46:47-0500] [ALPM] upgraded libmm-glib (1.14.6-1 -> 1.14.8-1)
[2020-11-21T07:46:48-0500] [ALPM] upgraded linux (5.9.8.arch1-1 -> 5.9.9.arch1-1)
[2020-11-21T07:46:50-0500] [ALPM] upgraded linux-docs (5.9.8.arch1-1 -> 5.9.9.arch1-1)
[2020-11-21T07:46:52-0500] [ALPM] upgraded linux-firmware (20201023.dae4b4c-1 -> 20201113.2ea8667-1)
[2020-11-21T07:46:56-0500] [ALPM] upgraded linux-headers (5.9.8.arch1-1 -> 5.9.9.arch1-1)
[2020-11-21T07:46:56-0500] [ALPM] upgraded lsof (4.93.2-3 -> 4.94.0-1)
[2020-11-21T07:46:56-0500] [ALPM] upgraded modemmanager (1.14.6-1 -> 1.14.8-1)
[2020-11-21T07:46:56-0500] [ALPM] upgraded openblas (0.3.10-1 -> 0.3.12-1)
[2020-11-21T07:46:56-0500] [ALPM] upgraded openlibm (0.7.2-1 -> 0.7.3-1)
[2020-11-21T07:46:57-0500] [ALPM] upgraded oxygen-icons (1:5.75.0-1 -> 1:5.76.0-1)
[2020-11-21T07:46:57-0500] [ALPM] upgraded perl-file-listing (6.04-8 -> 6.11-1)
[2020-11-21T07:46:57-0500] [ALPM] upgraded perl-uri (1.76-4 -> 5.05-1)
[2020-11-21T07:46:57-0500] [ALPM] upgraded perl-http-cookies (6.08-3 -> 6.09-1)
[2020-11-21T07:46:57-0500] [ALPM] upgraded prison (5.75.0-1 -> 5.76.0-1)
[2020-11-21T07:46:57-0500] [ALPM] upgraded pulseaudio (13.99.3-1 -> 13.99.3+4+gd83ad6990-1)
[2020-11-21T07:46:57-0500] [ALPM] upgraded pulseaudio-bluetooth (13.99.3-1 -> 13.99.3+4+gd83ad6990-1)
[2020-11-21T07:46:57-0500] [ALPM] upgraded python-dnspython (2.0.0-1 -> 1:1.16.0-1)
[2020-11-21T07:46:57-0500] [ALPM] upgraded python-markdown (3.3-1 -> 3.3.3-1)
[2020-11-21T07:46:57-0500] [ALPM] upgraded python-pep517 (0.8.2-1 -> 0.9.1-1)
[2020-11-21T07:46:57-0500] [ALPM] upgraded python-resolvelib (0.4.0-1 -> 0.5.1-1)
[2020-11-21T07:46:57-0500] [ALPM] installed python-pyopenssl (19.1.0-2)
[2020-11-21T07:46:57-0500] [ALPM] upgraded python-pip (20.1.1-1 -> 20.2-1)
[2020-11-21T07:46:57-0500] [ALPM] upgraded python-regex (2020.10.28-1 -> 2020.11.13-1)
[2020-11-21T07:46:57-0500] [ALPM] upgraded python2-pep517 (0.8.2-1 -> 0.9.1-1)
[2020-11-21T07:46:57-0500] [ALPM] upgraded python2-resolvelib (0.4.0-1 -> 0.5.1-1)
[2020-11-21T07:46:57-0500] [ALPM] upgraded python2-pip (20.1.1-1 -> 20.2-1)
[2020-11-21T07:46:57-0500] [ALPM] upgraded syncthing (1.10.0-1 -> 1.11.1-1)
[2020-11-21T07:46:58-0500] [ALPM] upgraded syntax-highlighting (5.75.0-1 -> 5.76.0-1)
[2020-11-21T07:46:58-0500] [ALPM] upgraded system-config-printer (1.5.13-1 -> 1.5.13-3)
[2020-11-21T07:46:58-0500] [ALPM] installed botan (2.17.2-1)
[2020-11-21T07:46:58-0500] [ALPM] upgraded thunderbird (68.12.0-1 -> 78.4.3-1)
[2020-11-21T07:46:58-0500] [ALPM] upgraded tpm2-tss (3.0.1-1 -> 3.0.2-2)
[2020-11-21T07:46:58-0500] [ALPM] upgraded usbutils (012-2 -> 013-1)
[2020-11-21T07:46:58-0500] [ALPM] upgraded vala (0.50.1-1 -> 0.50.2-1)
[2020-11-21T07:46:59-0500] [ALPM] upgraded youtube-dl (2020.11.01.1-1 -> 2020.11.19-1)
[2020-11-21T07:46:59-0500] [ALPM] transaction completed

Offline

#34 2020-11-24 06:33:29

Salkay
Member
Registered: 2014-05-22
Posts: 618

Re: [FIXED?] Chromium artefacts after resume from suspend or tty switch

eblau wrote:

After updating two days back, this problem has gone away for me with chromium.

FWIW this is in no way fixed for me, testing with a tty switch.

Offline

#35 2020-11-25 13:34:44

eblau
Member
Registered: 2016-02-11
Posts: 43

Re: [FIXED?] Chromium artefacts after resume from suspend or tty switch

Salkay wrote:
eblau wrote:

After updating two days back, this problem has gone away for me with chromium.

FWIW this is in no way fixed for me, testing with a tty switch.

tty switch works fine for me now and I've suspended and resumed successfully with chromium hardware acceleration at least half a dozen times now. Prior to updating, I had artifacts displayed in chromium every single time.

Before the latest set of updates, I tried enabling vulkan but chromium slowed down so much as to make my entire system unusable. I was lucky to be able to even disable it in chrome://flags. Performance was horrible, at least with my laptop with NVIDIA graphics and dual 4K displays.

Offline

#36 2021-03-04 22:52:38

Neven
Member
Registered: 2014-05-02
Posts: 74

Re: [FIXED?] Chromium artefacts after resume from suspend or tty switch

Hey people. It seems like a lot of you are/were having trouble with the Wiki instructions for NVIDIA's experimental power management features. I'm the one who wrote the section (and I now clarified it a bit, hopefully) and my setup works perfectly for me, so I'd like to help - just ask away.

The summary for what needs to be done (in a "cookbook" style) is:

0. Find a filesystem with enough free space. From now on, I'm assuming this filesystem is mounted at

/

1. Make the directory

/tmp-nvidia

2. Put

options nvidia NVreg_PreserveVideoMemoryAllocations=1 NVreg_TemporaryFilePath=/tmp-nvidia

into

/etc/modprobe.d/nvidia-power-management.conf

3. Enable the

nvidia-suspend

and/or

nvidia-hibernate

Systemd services.

4. Reboot

BTW, the Wiki guys have for some reason decided to hide the section, so now it can be found here: https://wiki.archlinux.org/index.php/NV … er_suspend

Also note that the folks at NVIDIA have now added another completely different experimental system for safe power management which doesn't depend on a filesystem or a disk, see their documentation (linked from wiki) for that. The docs are also present in /usr/share/doc/nvidia/html/powermanagement.html in the nvidia-utils package.

Last edited by Neven (2021-03-04 23:06:17)

Offline

#37 2021-03-05 02:27:37

Salkay
Member
Registered: 2014-05-22
Posts: 618

Re: [FIXED?] Chromium artefacts after resume from suspend or tty switch

Thanks for the writeup @Neven, but I'm not sure if it's relevant to the TTY-switch artefacts?

Offline

#38 2021-03-05 06:44:47

kokoko3k
Member
Registered: 2008-11-14
Posts: 2,390

Re: [FIXED?] Chromium artefacts after resume from suspend or tty switch

Salkay, unfortunately we're out of luck there.
tty switch is broken (and so is broken even a basic multi-user and graphical system) on proprietary nvidia drivers.
I don't know of any working hack around that other than using an intel igp as the primary card and nvidia via prime render offload.


Help me to improve ssh-rdp !
Retroarch User? Try my koko-aio shader !

Offline

#39 2021-03-05 06:59:39

Salkay
Member
Registered: 2014-05-22
Posts: 618

Re: [FIXED?] Chromium artefacts after resume from suspend or tty switch

Thanks for that info @kokoko3k. That's a bit annoying, but good to know it's not worth chasing too much.

I mainly use Firefox, so I won't worry fiddling around then.

Offline

#40 2021-03-06 00:23:17

Neven
Member
Registered: 2014-05-02
Posts: 74

Re: [FIXED?] Chromium artefacts after resume from suspend or tty switch

I suppose the NVIDIA power management features indeed have absolutely nothing to do with switching TTYs.

Last edited by Neven (2021-03-06 00:55:17)

Offline

#41 2021-03-06 07:07:05

kokoko3k
Member
Registered: 2008-11-14
Posts: 2,390

Re: [FIXED?] Chromium artefacts after resume from suspend or tty switch

I suspect they have instead, in the sense that those power management "features" are just ugly hacks around the main issue about bad video memory managment that affects suspension,hibernation and VT switching.

(speculating here)
Probably the kernel does not provide a callback feature for vt switch like it does for suspension and this makes impossible for the driver to kick in in time to save and/or restore video memory "on the fly".
...and even if it could, that would mean incredibly slow switch and a lot of disk space usage, for every opened VT.

I'm not sure this is what we want.

Last edited by kokoko3k (2021-03-06 07:10:41)


Help me to improve ssh-rdp !
Retroarch User? Try my koko-aio shader !

Offline

#42 2021-03-06 08:26:04

seth
Member
Registered: 2012-09-03
Posts: 50,012

Re: [FIXED?] Chromium artefacts after resume from suspend or tty switch

The nvidia blob doesn't support framebuffer consoles, try eg. https://wiki.archlinux.org/index.php/GR … ramebuffer
This problem is different from video memory preservation during an S3 and most of these problems boil down to nvidia trying to bend the kernel around there shared driver infrastructure and kernel devs in return GPL-blocking features away from the driver.

The germans have a word for that which the rest of the world borrows: Kindergarten.

Offline

#43 2021-03-07 06:06:24

Salkay
Member
Registered: 2014-05-22
Posts: 618

Re: [FIXED?] Chromium artefacts after resume from suspend or tty switch

Thanks for that info @seth. I was intending on testing some different options, but before I did that I wanted to confirm the issue was still present. I did a few suspend-resume cycles, and a few TTY switches, and it seems like cura is totally fine now. Not sure what changed. Anyway, I'll keep it in mind if I get problems again in the future. Thanks all.

Offline

Board footer

Powered by FluxBB