You are not logged in.
Hello! I own a NVIDIA GTX 1060 GPU and updated my system yesterday evening, most importantly: `linux 4.16.8-1` and `nvidia 396.24-2`. After booting today, X didn't start properly. I have a dual-monitor setup, one screen didn't seem to receive any HDMI signal, while the other one displayed a glitchy mess (horizontal white "snow" streaks). Figuring that it was probably the driver update, I chrooted into my system and installed the legacy NVIDIA drivers: `nvidia-390xx 390.48-2` and everything is fine again.
I couldn't find similar reports of this problem in this forum or the subreddit, which makes me afraid that I'm an insolated case. Has this happened to anyone else? I'll try to gather some logs related to this issue.
Offline
https://devtalk.nvidia.com/default/topi … ring-min-/
https://devtalk.nvidia.com/default/topi … 6-doesn-t/
Last edited by blablo (2018-05-11 13:14:29)
Offline
Dont worry, you are not an isolated case.
I'm using an Optimus Laptop (nvidia + intel)
I had to downgrade the Kernel to 4.16.7-1 + nvidia + nvidia-utils to make it work again.
1) cd /var/cache/pacman/pkg/
2) sudo pacman -U nvidia-390.48-13-x86_64.pkg.tar.xz nvidia-utils-390.48-1-x86_64.pkg.tar.xz nvidia-settings-390.48-1-x86_64.pkg.tar.xz
3) sudo pacman -U linux-4.16.7-1-x86_64.pkg.tar.xz
4) Reboot & profit, I guess.
Last edited by Misagi (2018-05-11 14:32:13)
Offline
lib32-nvidia-utils as well, probably.
Help me to improve ssh-rdp !
Retroarch User? Try my koko-aio shader !
Offline
FWIW instead of downgrading you could also switch to the 390xx branch (if you happen to have a Fermi card that "could" becomes a must)
Last edited by V1del (2018-05-12 09:17:31)
Offline
That's a relief, thanks guys! I'll keep running with the legacy 390xx drivers for now until I hear news about a fix.
Last edited by cryzed (2018-05-11 19:37:43)
Offline
You could also disable nvidia-drm.modeset, which worked for me on GTX 980 Ti. After all, it's not that you benefit a lot from nvidia modesetting at the current point, other than fan speed going down earlier and having the console across multiple screens.
Offline
try on nvidia-390.48 and kernel linux-4.16.8
and not working
Offline
Same problem with GTX 1080 eGPU. However, downgrading to the 390xx branch did not solve it for me. I noticed a warning about AllowExternalGpus in Xorg.0.log, but it still does not work with the following:
Section "Device"
Identifier "NVidia Graphics"
Driver "nvidia"
VendorName "NVIDIA Corporation"
Option "AllowExternalGpus" "on"
EndSection
I do not have nvidia-drm.modeset=1 in my kernel boot line, and I have WaylandEnabled=false in gdm.conf
Offline
I currently have linux 4.16.8-1 (custom build) + nvidia-dkms 396.24-2 running on a 1070 without any issues. On my work laptop I have a 1060 and it also runs without any issues. I'm actually pleasantly surprised that I haven't been hit with this bug. Heh. Nvidia is absolute mess when it comes to drivers. I'm thinking about going to iGPU since I don't play games and my Intel CPU supports 4K @ 60Hz. I'm sick of all these Nvidia issues. They can never get anything right on the first try and break installs all the damn time.
I wonder why some people have this issue and some don't. Previous Nvidia issue (post-Spectre driver mess) impacted everyone.
Last edited by Batou (2018-05-18 02:45:11)
Please vote for all the AUR packages you're using. You can mass-vote for all of them by doing: "pacman -Qqm | xargs aurvote -v" (make sure to run "aurvote --configure" first)
Offline
@Batou This was a handy reminder for me at least (Nvidia 1030, kernel ARCH). Thx!
Seeded last month: Arch 50 gig, derivatives 1 gig
Desktop @3.3GHz 8 gig RAM, linux-ck
laptop #1 Atom 2 gig RAM, Arch linux stock i686 (6H w/ 6yrs old battery ) #2: ARM Tegra K1, 4 gig RAM, ChrOS
Atom Z520 2 gig RAM, OMV (Debian 7) kernel 3.16 bpo on SDHC | PGP Key: 0xFF0157D9
Offline
i also had to downgrade to the nvidia 390xx packages, it worked until today, but after upgrading xorg to 1.20 it was broken again
had to downgrade xorg to 1.19 to make it work:
cd /var/cache/pacman/pkg/
pacman -U xorg-server-1.19.6+13+gd0d1a694f-2-x86_64.pkg.tar.xz xorg-server-common-1.19.6+13+gd0d1a694f-2-x86_64.pkg.tar.xz xorg-server-xvfb-1.19.6+13+gd0d1a694f-2-x86_64.pkg.tar.xz xorg-server-xwayland-1.19.6+13+gd0d1a694f-2-x86_64.pkg.tar.xz
it seems 390xx is not compatible with xorg 1.20:
[ 4.352] (II) LoadModule: "nvidia"
[ 4.352] (II) Loading /usr/lib/xorg/modules/drivers/nvidia_drv.so
[ 4.352] (II) Module nvidia: vendor="NVIDIA Corporation"
[ 4.352] compiled for 4.0.2, module version = 1.0.0
[ 4.352] Module class: X.Org Video Driver
[ 4.352] ================ WARNING WARNING WARNING WARNING ================
[ 4.352] This server has a video driver ABI version of 24.0 that this
driver does not officially support. Please check
http://www.nvidia.com/ for driver updates or downgrade to an X
server with a supported driver ABI.
[ 4.352] =================================================================
[ 4.352] (EE) NVIDIA: Use the -ignoreABI option to override this check.
[ 4.352] (II) UnloadModule: "nvidia"
[ 4.352] (II) Unloading nvidia
[ 4.352] (EE) Failed to load module "nvidia" (unknown error, 0)
Because we don't think about future generations, they will never forget us.
-- Henrik Tikkanen
Offline
I wonder why some people have this issue and some don't.
Offline
i also had to downgrade to the nvidia 390xx packages, it worked until today, but after upgrading xorg to 1.20 it was broken again
had to downgrade xorg to 1.19 to make it work:
Offline
Just like Batou above I have no issues running
linux-4.16.9-1-ARCH x86_64
xorg-server 1.20.0-2
nvidia-dkms 396.24-3
And then
nvidia-settings 396.24-1
nvidia-utils 396.24-1
opencl-nvidia 396.24-1
lib32-nvidia-utils 396.24-2
libvdpau 1.1.1+3+ga21bf7a-1
steam-1.0.0.54-3
On a GT 1030 ("Pascal" smthg).
I just uninstalled the old nvidia* pkgs then installed the new driver, starting with nvidia-dkms.
I have not tested Graysky's linux-ck yet.
Seeded last month: Arch 50 gig, derivatives 1 gig
Desktop @3.3GHz 8 gig RAM, linux-ck
laptop #1 Atom 2 gig RAM, Arch linux stock i686 (6H w/ 6yrs old battery ) #2: ARM Tegra K1, 4 gig RAM, ChrOS
Atom Z520 2 gig RAM, OMV (Debian 7) kernel 3.16 bpo on SDHC | PGP Key: 0xFF0157D9
Offline
Batou wrote:I wonder why some people have this issue and some don't.
Interesting.
It seems they've fixed the issue:
We have fixed this issue and the fix will be available in upcoming driver release ...
Please vote for all the AUR packages you're using. You can mass-vote for all of them by doing: "pacman -Qqm | xargs aurvote -v" (make sure to run "aurvote --configure" first)
Offline
I don't think NVIDIA has put out a release since they fixed this issue, correct? At least the 396.24 release still doesn't include it, as far as I can tell. This is a problem because I basically can't update my system at all, without running the risk of breaking something. Is there something that I don't know of, where I could tell when the next NVIDIA driver will be presumably released?
Offline
The nvidia-390xx branch should be updateable with the rest of the system, does that not work?
Last edited by V1del (2018-05-27 14:54:34)
Offline
No, it's apparently ABI incompatible with the shortly after released `xorg-server 1.20`, you can see this a few posts above. I had to downgrade `xf86-video-dummy`, `xf86-video-vesa`, `xorg-server` and `xorg-server-common` to get the previous 1.19 release working again and added them to my `IgnorePkg` array, just to be safe. However I think that updating my system in this state would definitely lead to issues.
EDIT: I've just taken a closer look at the linked threads, apparently 390.59 should work and is already in the repos. I'll try that out then, sorry.
EDIT 2: That worked without issues.
Last edited by cryzed (2018-05-27 18:08:31)
Offline
Thanks for this Post, i was totally lost yesterday after i've come back from vacation and did an update. I am currently running nouveau – not crazy bad – i didn't even thought about the 390xx legacy driver, nice to know. Looks like there is a 396.24-7 currently in the repos since yesterday but i didn't tried it yet. But what i read in the nvidia forum* linked above is, that the fix they did does not go into the 396 branch and it is recommended to stay on 390 until the next release? (or is it already in there, idk?)
*https://devtalk.nvidia.com/default/topic/1032725/linux/396-18-02-neon-sddm-crash-on-boot-xid-62-nvrm-rm_init_adapter-failed-for-device-bearing-min-/post/5262427/#5262427
Offline
Sorry for necrobumping of this topic. For anyone, who still has the problem (like I did) - solution was quite simple - switch from nvidia package to nvidia-dkms package. No downgrading is necessary. Current working package versions:
linux 4.17.4-1
xorg-server 1.20.0-9
nvidia-dkms 396.24-14
nvidia-settings 396.24-1
nvidia-utils 396.24-2
lib32-nvidia-utils 396.24-2
libvdpau 1.1.1+3+ga21bf7a-1
libxnvctrl 396.24-1
Offline