[SOLVED] startx hangs after kernel update

Avatar-J · 2019-05-29 23:14:10

Edit - marked as solved, removed blacklist intel from /etc/modprobe.d/blacklist.conf

Any kernel higher than 5.0.10 results in a system freeze after startx is run.
KDE fails to load and the system stays on tty1 not accepting any input.
Originally i didn't need to configure anything after installing drivers, the nvidia drivers were loaded and multi-screen worked fine.

Screenshot 01 - Very low resolution, previously the TTY was displayed at 3840 x 2160
https://imgur.com/13nexNY

Screenshot 02 - X just hangs after startx .
https://imgur.com/oY6eoeZ

[ System before the issues ]
OS: Arch Linux x86_64
Host: Alienware m17 1.5.2
Kernel: Linux 5.0.10-arch1-1-ARCH #1 SMP PREEMPT Sat Apr 27 20:06:45 UTC 2019 x86_64 GNU/Linux
Shell: bash 5.0.7
Resolution: 2880x1620 (Max resolution is 3840 x 2160)
DE: KDE
WM: KWin
CPU: Intel i9-8950HK (12) @ 4.800GHz
GPU: Intel UHD Graphics 630
GPU: NVIDIA GeForce RTX 2080 Mobile
Memory: 3661MiB / 31959MiB

[ GPUs ]
lspci -k | grep -A 2 -E "(VGA|3D)"
00:02.0 VGA compatible controller: Intel Corporation UHD Graphics 630 (Mobile)
DeviceName: Onboard - Video
Subsystem: Dell UHD Graphics 630 (Mobile)
--
01:00.0 VGA compatible controller: NVIDIA Corporation TU104M [GeForce RTX 2080 Mobile] (rev a1)
Subsystem: Dell TU104M [GeForce RTX 2080 Mobile]
Kernel driver in use: nvidia

[ Breaking the system ]
The system broke after updating to the following packages
core/linux-5.1.5.arch1-2
extra/libvdpau 1.2-1
extra/libxnvctrl 430.14-1
extra/nvidia 430.14-6
extra/nvidia-utils 430.14-1

I assumed (possibly incorrectly) that new versions of the nvidia driver or kernel would require some extra configuration to resolve this issue.
I did try to find anything recent and/or relevant on the bug tracker but didnt have any success.

ran nvidia-config - output file /etc/X11/xorg.conf - moved this and changed name to:
/etc/X11/xorg.conf.d/20-nvidia.conf

# nvidia-xconfig: X configuration file generated by nvidia-xconfig
# nvidia-xconfig:  version 430.14

Section "Module"
	Load "modesetting"
EndSection

Section "Device"
    Identifier     "nvidia"
    Driver         "nvidia"
    BusID	   "1:0:0"
    Option	   "AllowEmptyInitialConfig"
EndSection

Section "Screen"
    Identifier     "Screen0"
    Device         "Device0"
    Monitor        "Monitor0"
    DefaultDepth    24
    SubSection     "Display"
        Depth       24
    EndSubSection
EndSection

add lines to ~/.xinitrc

    xrandr --setprovideroutputsource modesetting NVIDIA-0
    xrandr --auto
    xrandr --dpi 96

    exec startkde

Modified my /etc/default/grub with

nvidia-drm.modeset=1

rebuilt grub and rebooted - issue persists

Modified grub again, adding a second option after nvidia-drm

nvidia-drm.modeset=1

rebuilt grub and rebooted - issue persists

Previously I did not need to add anything to mkinitcpio.conf. I added these options:

MODULES(nvidia nvidia_modeset nvidia_uvm nvidia_drm)

Then rebuilt the initramfs

 mkinitcpio -p linux

Lastly I added the following option to /etc/default/grub

rcutree.rcu_idle_gp_delay=1

(added before nvidia-drm.modeset=1)
Then rebuilt my grub config.

I've had a look through what i think are relevant log files, in some of them i could not see anything i thought was relevant.

Links to log files
/etc/X11/xorg.conf.d/20-nvidia.conf = https://www.dropbox.com/s/p5wph8wsde58v … .conf?dl=0
~/.local/share/xorg/Xorg.0.log = https://www.dropbox.com/s/pjqiduczgfw2y … g.log?dl=0
/var/log/Xorg.0.log = https://www.dropbox.com/s/hogx72d5ciqt4 … 0.log?dl=0
/var/log/Xorg.1.log = https://www.dropbox.com/s/vyxjbyw1d94k2 … 1.log?dl=0
/var/log/Xorg.0.log.old = https://www.dropbox.com/s/2dzrfp4r3j0w0 … g.old?dl=0
dmesg = https://www.dropbox.com/s/0j95edxtvxppp … utput?dl=0
journalctl -xe = https://www.dropbox.com/s/ds1fsni7kr5ql … utput?dl=0

I must be doing something wrong but cannot figure out what. Any help would to point me in the right direction would be greatly appreciated.

Last edited by Avatar-J (2019-06-02 18:42:30)

Kartious · 2019-05-30 10:50:04

I don't think I can add much to this. But a couple of things to get the ball rolling.

Can you switch tty's while X is hanging in img2? If you can, are you able to view the current output of the log?
Are you able to start a minimal clean X environment ? Such as twm?
I see you have dual GPU setup on a laptop, do you/ have you tried to configure Bumblebee?
If Bumblebee is a complete nono, have you tried using the nvidia-lts package? (and at a stretch the linux-lts kernel?)

Avatar-J · 2019-05-30 12:27:16

Hi Kartious

Yes i can switch to tty2 when this happens, i do this to end processes such as plasma and latte that eat up 100% of a single core each. They dont behave like this when KDE starts normally.

I installed openbox and xorg-twm to test other environments.
Open box hangs the same as plasma
TWM doesnt but the X server terminates and directs me to the following log
/var/log/Xorg.2.log = https://www.dropbox.com/s/9p614plhh7w7n … 2.log?dl=0

I've not tried bumblebee yet, my logic was not to move on and try configure something else when the current setup does not work.

loqs · 2019-05-30 12:42:16

If you remove the Xorg.conf and ~/.xinitrc changes then downgrad nvidia-utils to 418.74-1 and replace nvidia with nvidia-dkms 418.74-4 (ensuring linux-headers are installed so DKMS can generate the module) what is the result?

Avatar-J · 2019-05-30 13:55:50

Hi loqs

I moved the 20-nvidia.conf from /etc/X11/xorg.conf.d to a folder in home so its out of the way, then removed those extra xrandr arguments form ~/.xinitrc.
I then installed linux-headers
I dont have nvidia-dkms 418.74-4 in my package cache.

I tried to download and modify the PKGBUILD for nvidia-dkms 430.14-6 but makepkg was unable to download the source for this.

Being new to arch and building packages, i found this link:
https://git.archlinux.org/svntogit/pack … 39a0e39db5

Maybe i can clone the source from the git link in that page.
I'm gonna read up on creating packages and pkgbuilds and see if i can build this.
.

Lone_Wolf · 2019-05-30 14:01:04

You can probably find that package in the Arch LInux Archive

Last edited by Lone_Wolf (2019-05-30 14:01:18)

Avatar-J · 2019-05-30 15:18:17

Thank you Lone_Wolf and loqs, I had no idea the archive existed.
I was able to download and install nvidia-dkms 418.74-4 and nvidia-utils-418.74.-4

The result is the same after rebooting the system.
startx said it logged events to ~/.local/share/xorg/Xorg.0.conf
A copy of that log is here = https://www.dropbox.com/s/2dzrfp4r3j0w0 … g.old?dl=0

Googling one error from the log "Failed to allocate software rendering cache surface: out of memory", yielded the following result.
https://www.astroarch.com/2019/01/nvidi … e-surface/

In this link, the author removes "nopat" from kernel arguments in the bootloader, i did the same, rebuilt the grub and rebooted.
X fails and produces this log = https://www.dropbox.com/s/c8rvcp1es1pl7 … nopat?dl=0
I had a look through the log, i dont think any of the EE entries relate to the nvidia driver.
Theres a few WW and II entries that reference "unloading modesetting" and "Please see the ConnectToAcpid X configuration option"

loqs · 2019-05-30 15:29:15

The latest Xorg.logs look like an unconfigured Optimus system.
If you set Xorg.conf to contain

Section "Module"
	Load "modesetting"
EndSection

Section "Device"
    Identifier     "nvidia"
    Driver         "nvidia"
    BusID	   "1:0:0"
    Option	   "AllowEmptyInitialConfig"
EndSection

and add the following back to ~/.xinitrc

    xrandr --setprovideroutputsource modesetting NVIDIA-0
    xrandr --auto
    xrandr --dpi 96

Is there any change?

Avatar-J · 2019-05-30 15:45:33

I tried the above suggestions but unfortunately the issue persists.

seth · 2019-05-30 15:48:26

tried

Updated logs and xinitrc?

loqs · 2019-05-30 15:50:24

Please post the contents of /etc/X11/xorg.conf and /etc/X11/*.conf files and ~/.xinitrc as they are now as well as the latest Xorg.log.

Avatar-J · 2019-05-30 16:15:25

I looked in /etc/X11 for any .conf files, the only one is the following:
/etc/X11/xorg.conf.d/20-nvidia.conf = https://www.dropbox.com/s/erd01a2kr96ew … .conf?dl=0

~/.xinitrc = https://www.dropbox.com/s/lnzdutdqowk4n2u/xinitrc?dl=0

I tried moving and renaming the 20-nvidia.conf to /etc/X11/xorg.conf but the issue remained

The most current log is:
~/.local/share/xorg/Xorg.0.log = https://www.dropbox.com/s/jd4gm4jicbh10 … 0.log?dl=0

I really appreciate all the help from you folks.

seth · 2019-05-30 16:24:31

Can you please stop using dropbox and some pastebin service? It sucks to have to download every file because DP cannot display textfiles unless they end on .txt.

"exec xorg-twm" will not do anything and the server terminate quickly. The binary is just "twm" and you need to have it installed itfp.

Avatar-J · 2019-05-31 11:08:47

Twm is installed as per comment number 3.
I'll look again today for any bugs and at the changelog for kernel 5.0.11 to see if I can find anything

seth · 2019-05-31 12:41:30

You're still invoking it wrongly which is why the logs reflect and insta-close, so right now we've got a plasma thing, the flawed twm log and an assessment that openbox hangs "like plasma" (though from the screenshot it looks less like "hanging" and more like "VT doesn change and isn't occupied by X11 either)
And a broken xinitrc (apparently you've been using that all along? Please read the xinit wiki)

Also I downloaded more files from the [veryuglyswearinghere] dropbox, the dmesg (like the Xorg logs) shows no trace of the intel IGP, but there seem no outputs wired to the nvidia GPU. That's a problem. What happened to the chip? Is the i915 module loaded (the lspci excerpt does not assert that)?

You also may want to try to enforce X11 to run as root: https://wiki.archlinux.org/index.php/Xorg#Rootless_Xorg

Avatar-J · 2019-06-02 18:41:05

So i read the xinit wiki page, replaced the broken ~/xinitrc with the example one and then added the xrandr and kde things.
Then following your advice i searched lsmod for i915 and intel, looking at the output there was no intel graphics driver being loaded.
I checked /etc/modprobe.d/blacklist.conf and in this file both Intel and nouveau were set.
I removed blacklist intel and rebooted.

Now everything works fine, i am on the latest kernel and latest nvidia-dkms and nvidia-utils packages, the intel and nvidia drivers load fine.
Thank you for all your help, and in future i will use pastebin

Arch Linux

#1 2019-05-29 23:14:10

[SOLVED] startx hangs after kernel update

#2 2019-05-30 10:50:04

Re: [SOLVED] startx hangs after kernel update

#3 2019-05-30 12:27:16

Re: [SOLVED] startx hangs after kernel update

#4 2019-05-30 12:42:16

Re: [SOLVED] startx hangs after kernel update

#5 2019-05-30 13:55:50

Re: [SOLVED] startx hangs after kernel update

#6 2019-05-30 14:01:04

Re: [SOLVED] startx hangs after kernel update

#7 2019-05-30 15:18:17

Re: [SOLVED] startx hangs after kernel update

#8 2019-05-30 15:29:15

Re: [SOLVED] startx hangs after kernel update

#9 2019-05-30 15:45:33

Re: [SOLVED] startx hangs after kernel update

#10 2019-05-30 15:48:26

Re: [SOLVED] startx hangs after kernel update

#11 2019-05-30 15:50:24

Re: [SOLVED] startx hangs after kernel update

#12 2019-05-30 16:15:25

Re: [SOLVED] startx hangs after kernel update

#13 2019-05-30 16:24:31

Re: [SOLVED] startx hangs after kernel update

#14 2019-05-31 11:08:47

Re: [SOLVED] startx hangs after kernel update

#15 2019-05-31 12:41:30

Re: [SOLVED] startx hangs after kernel update

#16 2019-06-02 18:41:05

Re: [SOLVED] startx hangs after kernel update

Board footer