Black screen and freeze on boot with kernel 7.0.3 and Nvidia GPU

[banger696-prog]

Please confirm there isn't an open report for this issue

• I have searched through bug reports

Summary

Cant boot into Kernel with standard command line but adding nomodeset to kernel command line allows boot. See attached boot log in text format.

boot.log

Steps to reproduce

1. Sync with current Polaris repo.
2. Attempt reboot.
3. Freezes after black screen with cursor.

Expected result

Normal boot (which can be achieved with nomodeset in kernel command line in GRUB)

Actual result

Black screen and machine freeze

Environment

• Is system up to date?

Repo

Polaris (stable)

Desktop Environment

Plasma

System details

System:
Host: tim-solus Kernel: 7.0.3-336.current arch: x86_64 bits: 64
Desktop: KDE Plasma v: 6.6.4 Distro: Solus 4.9 serenity
Machine:
Type: Desktop Mobo: MICRO-STAR model: MS-7360 v: 1.0
serial: Firmware: BIOS vendor: American Megatrends
v: 1.10 date: 11/11/2008
CPU:
Info: dual core Intel Core2 Duo E8400 [MCP] speed (MHz): avg: 3004
Graphics:
Device-1: NVIDIA TU117 [GeForce GTX 1650] driver: nvidia v: 580.142
Display: wayland server: X.org v: 1.21.1.22 with: Xwayland v: 24.1.10
compositor: kwin_wayland driver: gpu: nv_platform,nvidia,nvidia-nvswitch
resolution: 1920x1080~60Hz
API: OpenGL v: 4.6.0 compat-v: 4.5 vendor: nvidia mesa v: 580.142
renderer: NVIDIA GeForce GTX 1650/PCIe/SSE2
Info: Tools: api: clinfo, eglinfo, glxinfo, vulkaninfo
de: kscreen-console,kscreen-doctor gpu: lact, nvidia-settings, nvidia-smi
wl: wayland-info x11: xdpyinfo, xprop, xrandr
Network:
Device-1: Realtek RTL8111/8168/8211/8411 PCI Express Gigabit Ethernet
driver: r8169
Device-2: Realtek RTL88x2bu [AC1200 Techkey] driver: rtw88_8822bu
type: USB
Drives:
Local Storage: total: 609.23 GiB used: 353.41 GiB (58.0%)
Info:
Memory: total: 8 GiB available: 7.75 GiB used: 2.9 GiB (37.4%)
Processes: 263 Uptime: 14m Shell: Bash inxi: 3.3.40

Other comments

No response

[banger696-prog]

Also adding modprobe.blacklist=nouveau allows boot as well as nomodeset in kernel grub command line.

[HarveyDevel]

Also adding modprobe.blacklist=nouveau allows boot as well as nomodeset in kernel grub command line.

If blacklisting nouveau on its own fixes it, then it seems to be an issue specific to grub installs picking up the blacklist because its already blacklisted:

21e856c#diff-4cd13c585f04ebb6589a544c791fc26b995c1f587206aa307c856f82d4c1b91c

[ReillyBrogan]

Does our grub logic pull in the initrd extension? @silkeh @EbonJaeger ?

[banger696-prog]

I was browsing the boot log and there was an entry shortly before freeze that the nouveau had take control and nvidia couldn't load.

May 04 03:15:39 tim-solus kernel: NVRM: The NVIDIA probe routine was not called for 1 device(s).
May 04 03:15:39 tim-solus kernel: NVRM: This can occur when another driver was loaded and 
                                  NVRM: obtained ownership of the NVIDIA device(s).
May 04 03:15:39 tim-solus kernel: NVRM: Try unloading the conflicting kernel module (and/or
                                  NVRM: reconfigure your kernel without the conflicting
                                  NVRM: driver(s)), then try loading the NVIDIA kernel module
                                  NVRM: again.
May 04 03:15:39 tim-solus kernel: NVRM: No NVIDIA devices probed.

[infinitymdm]

I had this issue with a GTX 1650 Super after updating to the new kernels. Previous boot logs have a handful of errors along the lines of NVRM: GPU XXXX:XX:XX.X is already bound to nouveau. Adding a .conf file in /etc/modprobe.d containing options nouveau modeset=0 allowed my system to boot once more.