[Testing Update] 2022-01-03 - OpenSearch 1.2.3, Nvidia 495.46, Wine 7.0-rc3, LibAdwaita 1.0, Mesa 21.3.3

I assume you mean NVIDIA-390? If so that is expected as NVIDIA is supporting that till the end of 2022.

Yes, indeed, I meant 390. This keeps happening to me every time I type that version number :frowning:

:stop_sign: Just FYI for anyone using optimus-manager which is python package.

If you do not read updates, like me properly rebuild the python packages then your graphical interface, lightdm in my case, won’t start as it will repeatedly keep restarting, also taking away your tty control.

I had to ssh from phone to stop it and reinstall optimus-manager/rebuild python packages.

Happy new year people :tada:

There is only llvm-rc-bin in the AUR which should not be necessary. Remove it or install llvm from the repos instead.

1 Like

In the logs I could see a problem with the Nouveau driver. I installed the nonfree Nvidia driver instead and my GDM works again.

After enabling nouveau drivers I can’t get wayland back to work…

journalctl -p 3 -xb gives me these failures that look related…

gnome-session-binary[585]: Unrecoverable failure in required component org.gnome.Shell.desktop
gdm-launch-environment][551]: GLib-GObject: g_object_unref: assertion 'G_IS_OBJECT (object)' >

Seem to an issue related to OpenGL, after reinstalling mesa and lib32-mesa I could login under wayland again.

the first time you pushed nvidia 495.46, it ran fine for me. then you bumped it down to .44 and back up to .46 again today, and this time it’s not fine. running glxgears got me

BadValue (integer parameter out of range for operation)

so i downgraded to .44

and all is well again

Apparently opensearch and opensearch-cli are in conflict. Is that normal, given that opensearch-cli is listed as an optionnal dependency for opensearch?

You could/should ask the arch package maintainer: hashworks@archlinux.org

Sure. I’m reporting it here though, because that’s the testing branch.

Arch bug report submitted: FS#73256 - [opensearch] Conflicts with opensearch-cli

1 Like

i get a kernel panic under virtualbox , do not detect smtp processor any more
see this ( inxi under vbox )

info: model: AMD Ryzen 5 5600x bits :64 type : MCP arch : Zen 3 
Family 0x19 (25) modele-id : 0x21 (33) stepping :0 microcode:0x6000626
Topology cpus:1x cores:2 smt:<unsupported> etc..

A post was split to a new topic: Connecting my second monitor causes my desktop to crash

I am unable to find corresponding packages, like nvidia-utils-495xx, or nvidia-495xx-dkms, in the repository. What am I doing wrong?

The current NVIDIA driver does not have a version number in the name so it’s just nvidia-utils, nvidia-dkms, etc.

Thanks a lot for that update. This now seems to go to 495.46. Unfortunately, that version seems to be incompatible with kernel 5.16 yet:

==> dkms install --no-depmod nvidia/495.46 -k 5.16.0-1-MANJARO
Error! Bad return status for module build on kernel: 5.16.0-1-MANJARO (x86_64)
Consult /var/lib/dkms/nvidia/495.46/build/make.log for more information.
==> WARNING: `dkms install --no-depmod nvidia/495.46 -k 5.16.0-1-MANJARO' exited 10

UPDATE: For this problem, it looks like there is already a solution.
See mm/migrate.c: remove MIGRATE_PFN_LOCKED · torvalds/linux@ab09243 · GitHub for details.

As described there, the driver needs a small patch:

--- a/kernel/nvidia-uvm/uvm_migrate_pageable.c	2022-01-08 21:51:32.672454683 +0100
+++ b/kernel/nvidia-uvm/uvm_migrate_pageable.c	2022-01-08 21:51:52.298644944 +0100
@@ -406,7 +406,7 @@
         uvm_push_set_flag(&push, UVM_PUSH_FLAG_CE_NEXT_MEMBAR_NONE);
         copying_gpu->parent->ce_hal->memset_8(&push, dst_address, 0, PAGE_SIZE);
-        dst[i] = migrate_pfn(page_to_pfn(dst_page)) | MIGRATE_PFN_LOCKED;
+        dst[i] = migrate_pfn(page_to_pfn(dst_page));
     if (copying_gpu) {
@@ -490,7 +490,7 @@
         uvm_push_set_flag(&push, UVM_PUSH_FLAG_CE_NEXT_MEMBAR_NONE);
         copying_gpu->parent->ce_hal->memcopy(&push, dst_address, src_address, PAGE_SIZE);
-        dst[i] = migrate_pfn(page_to_pfn(dst_page)) | MIGRATE_PFN_LOCKED;
+        dst[i] = migrate_pfn(page_to_pfn(dst_page));
     // TODO: Bug 1766424: If the destination is a GPU and the copy was done by

But, with 496.46, the screen, especially in the terminal, starts to flicker, and sometimes, the terminal output even jumps up and down by one line, rendering the terminal almost unusable.

UPDATE This seems to be connected to the compositor. I am using picom here - as soon as I kill it, that jumping and flickering disappears.

And, my second screen gets disabled on every start.
Also, in contrast to former versions, I now have to start nvidia-settings with sudo to be able to apply any changes to display settings.
Unfortunately, even when I request the changes to be saved in the X configuration file, they do not survive a reboot - the second screen then gets disabled again.

Saving the nvidia-settings configuration and running nvidia-settings --load-config-only on login does not help - it looks like it is doing nothing.

Using EndeavourOS, 495.46 kept me from logging in, just returned me to the display manager. I ended up using the downgrade command to revert back to 495.44. Did this happen to anyone?

Are you lost? :thinking:

1 Like

yes to having to downgrade, though my issue was not that severe

I’m with Yochanan, are you lost?

1 Like

oldrocker99’s point could be to compare driver work under couple different OS environments. So I think they has at least that point.
Sometimes somebody get lost, sometimes others lacks to get payload idea of question. Payload for Manjaro could be comparison of work under another OS and oldrocker99 and

clarifies for us that it is a broad problem, not only Manjaro.
I think it is good to have such feedback also - to know is it a distros wide issue or Manjaro only: that info could point any user/dev to a possible issue solution which was applied under another environment to fix possible issues or to realize that it could be a package problem on GNU/Linux (or Arch-family distros).

1 Like