Are the nvidia modules included in the initrd?

Does the current NVIDIA driver even work with kernel 5.9?

https://forums.developer.nvidia.com/t/nvidia-driver-not-yet-supported-for-linux-kernel-5-9/157263

Apparently they aren't. I looked at /etc/mkinitcpio.conf and there is nothing related.
In the past I had problems with the Nvidia drivers but it never happened that the machine was completely inaccessible. I has been always able to login in text mode. Now the machine is apparently loaded (fan spins loud) but frozen.
I tried to downgrade by booting the usb flash, then chrooting to the mounted root partition and downgrading to the previous kernel (and everything related) version with pacman -U. Surprisingly the outcome with the downgraded kernel was the same. So maybe the problem is in something used to build the kernel. I was able to boot the machine by copying the kernel and the initramfs* files from the /boot of a virtual machine I had on another PC. Surprisingly again, for me, this is working fine.

Thanks @AK, interesting link. So we must wait a month before upgrading the kernel.
At this point I'm somewhat surprised being apparently only one of the few that is complaining about this problem
and by the lack of any warning on arch's home page.

@jamp have a look at  FS#68312  the symptoms do not match what you are reporting. If you blacklist the nvidia modules (nvidia nvidia-drm nvidia-modeset nvidia-uvm) or remove the package supplying those modules can you then boot a 5.9 kernel?

@loqs I blacklisted the modules and upgraded again the kernel (blacklisting tested on the working system, before upgrading to 5.9.1) and the symptoms are as reported before so it is not a problem due to NVidia modules.
Booting 5.9.1 does not leave apparently any notice in the logs.
My system is an Asus N750JV.

Thank you

@loqs I upgraded the virtual machine from which I took the 5.8.14 kernel to 5.9.1, then I "implanted" its kernel on the physical machine and now the 5.9.1 is running also there. So the problem should be in the upgrade process.

I noticed that the 5.9.1 initramfs* files generated by pacman on the physical machine are far smaller than the corresponding files generated by pacman on the VM

Physical

-rwxr-xr-x 1 root root 24413951 Oct 22 09:52 initramfs-linux-fallback.img
-rwxr-xr-x 1 root root 2126369 Oct 22 09:52 initramfs-linux.img
-rwxr-xr-x 1 root root 8958080 Oct 22 09:52 vmlinuz-linux


VM

-rw-r--r-- 1 root root 32133661 Oct 22 09:43 initramfs-linux-fallback.img
-rw-r--r-- 1 root root 10011765 Oct 22 09:43 initramfs-linux.img
-rw-r--r-- 1 root root 8958080 Oct 22 09:43 vmlinuz-linux

With the latter files the 5.9.1 kernel and nvidia modules run fine also on the physical machine

I upgraded to v 5.9.4 today and with my machine the problem is the same, i.e. the initramfs files remain much smaller than the ones I get upgrading another system (a virtual machine, but this should not be relevant).
With the shorter files the machine hangs at boot, then I take the files from the VM and put them on the Asus laptop and it starts fine.
I checked the various mkinitcpio.conf and related files on both machines but they look the same.
This problem happened upgrading from 5.8.14 to 5.9.1 and is still there.
Is there someone that could give advice about this problem?
Thank you

I finally found the reason of the problem. Something has removed the symlink lib64 -> lib from the /usr directory.
The /usr/bin/ldd script uses the /usr/lib64/ld-linux-x86-64.so.2 to find the dependencies of some executables that the mkinitcpio program stuffs into the initramfs files. Not finding the ld-linux-x86-64.so.2 utility the executables were reported as statically linked by ldd and the necessary shared libraries were not included in the final initramfs files.
I haven't yet found what happened and hope that was a faulty package that removed the symlink because I try to pay the maximun attention to the integrity of the system files and directories.