You are not logged in.
Salutations. Throughout my time using Arch, I have had a persistent problem that happens after I install Nvidia drivers through pacman, following the steps listed on the wiki page NVIDIA.
I have an NVIDIA GTX 1070 graphics card. This uses the Pascal architecture, NV130, which released between NV110, Maxwell; and NV190, Lovelace. According to the chart on the wiki page, this means I should use the packages nvidia, nvidia-lts, or nvidia-dkms.
I have tried nvidia and nvidia-dkms each. I also downloaded nvidia-utils, nvidia-settings, and installed the necessary packages for Xorg.
I removed all instances of 'kms' in /etc/mkinitcpio.conf (wiki says to do so from the HOOKS array, but I'm not clear on what we're doing here), regenerated the initial ramdisk with 'mkinitcpio -P', and after running nvidia-xconfig, rebooted. Now, every time I reboot, the system freezes on a message in the form:
Nvidia drivers from pacman cause failure to boot, freezing at "/dev/[root partition]: clean, [a]/[\b] files, [c]/[d] blocks"
Any ideas? In the past my workaround has been installing drivers straight from NVIDIA's website, but I know that this is not recommended.
Last edited by pooplinux (2025-06-10 13:54:37)
Offline
Can you still boot the multi-user.target (2nd link below) - in doubt along "nomodeset"?
Othwise, unless you can ssh into the system) reboot it by frenetically pressing ctrl+alt+del or the https://wiki.archlinux.org/title/Keyboa … el_(SysRq) (needs to be explicitly enabled at the bootloader!) and access the journal from some live distr or the install iso, https://wiki.archlinux.org/title/System … al_to_view
Please post your complete system journal for the current (or previous "… -b -1 …") boot:
sudo journalctl -b | curl -F 'file=@-' 0x0.st
installing drivers straight from NVIDIA's website, but I know that this is not recommended
That's an understatement. "Dumbfuck idea" is a better description
Offline
Are you sure it "freezes"?
You may have multiple issues, but for certain your Pascal GPU will require the following kernel parameters applied:
nvidia_drm.modeset=1 nvidia_drm.fbdev=0
See:
Offline
Since adding the kernel parameters in tekstryder's post, I have been able to pass the "clean, a/b files... etc" message by waiting a few minutes until it throws an error, which flashes by in a fraction of a second before it is replaced by the login screen.
Now, though I can get into the system without using SysRq, running nvidia-xconfig or startx results in all monitors going dark and reading 'no signal'. The former produces yet another too-fast-to-read error message.
Here's a system log for a previous boot. On the user end, I waited for a login screen, logged in, ran nvidia-xconfig, the screen went dark, and I hit 'reset' on my PC case.
0x0 is very nifty. Thanks for introducing me to it, seth!
Offline
running nvidia-xconfig
Don't. Remove the generated /etc/X11/xorg.conf and then please also post your Xorg log, https://wiki.archlinux.org/title/Xorg#General
Also maybe remove the "quiet" parameter for the time being (because of the boot messages, it will not fix anything)
Offline
I removed all instances of 'kms' in /etc/mkinitcpio.conf (wiki says to do so from the HOOKS array, but I'm not clear on what we're doing here), regenerated the initial ramdisk with 'mkinitcpio -P', and after running nvidia-xconfig, rebooted.
You removed the kms hook, but have you tried adding the nvidia modules to the initramfs?
See this: https://wiki.archlinux.org/title/NVIDIA#Early_loading
Offline
Thanks all. I removed the 'quiet' parameter from /etc/default/grub and regenerated the GRUB configuration file.
I added the nvidia, nvidia_modeset, nvidia_uvm, and nvidia_drm modules to the MODULES array in /etc/mkinitcpio.conf, and regenerated the initramfs.
Here is my Xorg log.
Offline
You're using an ISO? With a kernel from last Sept? What in the world is this?
Offline
[ 853.853] (==) Using config file: "/etc/X11/xorg.conf"
You're still enforcing the config, remove that file.
[ 853.885] (EE) NVIDIA: Failed to initialize the NVIDIA kernel module. Please see the
[ 853.885] (EE) NVIDIA: system's kernel log for additional error messages and
[ 853.885] (EE) NVIDIA: consult the NVIDIA README for details.
lsmod | grep -E '(nvidia|nouveau)'
modinfo nvidia
pacman -Qs nvidia
dkms status
With the geriatric kernel, you'll have to use dkms, it requires linux-headers (for your kernel) and IDK whether the 570xx kernel will actually build on it, so check/post the dkms build log.
That being said, you probably want to address #8 first before wasting too much effort on working around a questionable system setup.
Offline
You're using an ISO? With a kernel from last Sept? What in the world is this?
I gathered the Xorg log by chrooting into the system from the live medium. I thought the Xorg log would have been from the last time I actually used X, aka the last time I ran startx, which was when the system was booted from the actual install location.
You're still enforcing the config, remove that file.
Bizarre, I thought I did. I'll give it another go.
Offline
I thought the Xorg log would have been from the last time I actually used X, aka the last time I ran startx
It is. How was this system installed?
Offline
The log
[ 853.846] (==) Log file: "/var/log/Xorg.0.log", Time: Sat May 24 22:22:22 2025
also was a couple of days old at the time you were posting it (so it's possible that you removed the config, but have not been starting the X11 server since or were eg. using startx what will put the xorg log into ~/.local/share/xorg/ )
… did you maybe at some point try to startx as root from a chroot out of the install iso and the relevant logs actually *are* in ~/.local/share/xorg/ ?
Offline
/etc/X11/ still contains no config file. Only the xinit directory. I have no idea why a nonexistent config was still being enforced. [Edit: I think it's because the Xorg log is misleading; see the bolded line below.]
I would not be surprised if at some point, I foolishly tried to startx as root from a chroot out of the install iso. However ~/.local/share/Xorg.0.log is empty.
Scimmia, to my recollection I installed by following the Wiki's installation guide.
pacman -Syu successfully updated nvidia-dkms to 575.57.08. dkms status confirms it is installed.
linux-headers has been installed for a while now.
As a sanity check, here is an up-to-date system journal and Xorg log, produced from within the installed system.
System log: http://0x0.st/8gww.txt
Xorg log: http://0x0.st/8YV_.txt
What the hell? Despite having run startx many times since, the log still reads as being from May 24th, and displays my current system as the archiso. Installed is 6.14.10.arch1-1. It seems startx crashes my system (black screen, 'no signal') before even producing a log.
(On the off-chance it has any bearing on anything -- and I don't know why it would -- since my last post I disabled Trusted Platform Module via UEFI (technically switching from Firmware TPM to Discrete TPM) as mentioned here, as an alternative to masking the device in systemctl, in order to speed up booting.)
Last edited by pooplinux (2025-06-08 01:16:30)
Offline
[ 853.846] (==) Log file: "/var/log/Xorg.0.log", Time: Sat May 24 22:22:22 2025
Despite having run startx many times
That's not where xinit stores the log => ~/.local/share/xorg/
Offline
That's not where xinit stores the log => ~/.local/share/xorg/
However ~/.local/share/Xorg.0.log is empty.
Gotcha. The log file exists, but is empty, so there is nothing to share.
Last edited by pooplinux (2025-06-08 22:18:56)
Offline
startx > ~/wtf.startx 2>&1
and post ~/wtf.startx
Offline
The resulting file is sadly empty.
Offline
??
Is this after a reboot w/ the power button?
Do you get any output just running
/usr/bin/Xorg -nolisten tcp "$@" vt$XDG_VTNR
on the console login?
Offline
Great news! I tried a different monitor setup (both HDMI rather than one HDMI and one DVI), and now my monitors don't lose signal after running startx, and we get an actually meaningful log.
First off I was informed I forgot to install Xterm. Good Lord, that's embarassing. Took care of that, and now startx brings me into a cluster of three terminal windows against a black screen, as well as a mouse cursor. Hovering over the black regions of the screen turns the mouse cursor into an X. I have installed i3 Window Manager (i3-wm) but this arrangement isn't quite right -- the windows have no top bars to move them around, and I'm unable to use any mod key commands such as mod+shift+e, which should open a dialog to exit i3. I am able to open Firefox, but not to move the window around.
Latest Xorg log here.
Last edited by pooplinux (2025-06-10 02:41:36)
Offline
cluster of three terminal windows against a black screen, as well as a mouse cursor. Hovering over the black regions of the screen turns the mouse cursor into an X
Perfectly normal for the default xinitrc (if you install TWM, you also get a windowmanager, xclock will get you a clock…)
the windows have no top bars to move them around
No, i3 is a tiling WM
OPenbox or icewm might feel more familiar here.
The Xorg log looks unsuspicious, you can boot, you can launch X11 - so where are we at?
Offline
Yes, I've used i3 in the past; if memory serves right you can still click and drag windows by a top bar to move them to other monitors. I'll look into adjusting the configuration. Still unsure why mod key commands are unresponsive, but all this is a different issue so I'll mark this as solved.
Thanks for helping out a newbie! Cheers all.
Offline
https://i3wm.org/docs/userguide.html#_r … ng_exiting
About the titlebar, do you simply need to install a font? (iirc i3 doesn't drag one in)
Offline
User guide set me straight. All is well. Thanks seth!
Offline