[Testing Update] 2020-12-07 - BREAKAGE EXPECTED - Kernels, Mesa 20.3, Python 3.9, Pamac 10

If the packages are switched correctly (which they are not, check my post), then it will work just fine :man_shrugging:

Those have not actually been dropped, check the relevant post.

Poedit is broken since this update:

poedit
poedit: symbol lookup error: /usr/lib/liblucene++.so.0: undefined symbol: _ZN5boost16re_detail_10720019raise_runtime_errorERKSt13runtime_error

Had problems with Libreoffice too, but fixed those

With this “testing” update had problems with “Catfish” and “zeitgeist” as being in conflict …
Bluntly solved by removing " Catfish" for the time being …

maybe one of them needs rebuild against new python, that’s why they are in conflict

Hi, I understood that, but hardware detection selects video-linux instead of 390xx-bumblebee.

I had other things to do. Here is the missing part of switching the Nvidia drivers correctly:

@TotallyNotElite maybe review it.

1 Like

@TotallyNotElite maybe review it.

Will do.

@philm Tested, this does not work:

  1. manjaro-system script fails to copy the files, because they have not been installed yet. (Maybe this should be done in mhwd-db’s post update scripts?)
  2. Linux-latest still screws around with the nvidia packages
  3. After forcing the completion of this update by fixing the conflict, you end up with a franken-nvidia system:
    image
    This happens because nvidia-utils does not replace the previous nvidia-utils packages, unlike lib32-nvidia-utils.
    For some reason, lib32-nvidia-utils is now no longer marked as explicitly installed.
  4. It also seems like there are quite a few more packages that need to be dropped from the repo.

So in short, this is a complete mess right now.

on lock Xfce screen ,
lost info user connecting with password

no solution

I hesitated with the update and waited till today, because I have hybrid graphics with optimus-manager.

I did the update today, then recompiled optimus-manager-git, rebooted and everything seems to be OK.
It’s possible I will find some breakages later, but so far, the system boots correctly, everything loads, Steam works, I can switch between GPU modes, Steam is showing correct drivers (intel, noveau or Nvidia - depending on the GPU mode).

So far so good.

I had some info about the lack of the folder:

grep: /etc/X11/mhwd.d/nvidia.conf: No such file or directory

but that’s expected on a proper optimus-manager install. This didn’t create any problems so seems like it worked well anyway.

1 Like

I had same or similar issue on unstable (xfce) posted journal there. It was light-locker related. No fix was found by me. Wound up removing light-locker and installed xfce-4 screensaver and enabled lock screen with that. No issues with screensaver. This was on a notebook with external monitor as Primary (lid in down position, HDMI connection)

https://bugs.archlinux.org/task/68905

What I reported initially.

Poedit works again with todays update

something seems to be wrong with wofi_

Process 8196 (wofi) of user 60458 dumped core.

                                              Stack trace of thread 8201:
                                              #0  0x00007f026fb9da85 __strlen_avx2 (libc.so.6 + 0x162a85)
                                              #1  0x00007f026facab43 __strdup (libc.so.6 + 0x8fb43)
                                              #2  0x0000560274efb655 wofi_read_cache (wofi + 0xe655)
                                              #3  0x0000560274efe06d wofi_drun_init (wofi + 0x1106d)
                                              #4  0x0000560274efacf4 n/a (wofi + 0xdcf4)
                                              #5  0x0000560274efaee0 n/a (wofi + 0xdee0)
                                              #6  0x00007f026fc0d3e9 start_thread (libpthread.so.0 + 0x93e9)
                                              #7  0x00007f026fb3b293 __clone (libc.so.6 + 0x100293)

@philm

I can rollback to pre “BREAKAGE EXPECTED” state if you provide a new update for Nvidia to test the update process. Currently main issue is that the driver is not replaced properly only the lib32 driver is replaced.

After updating and reboot, the display-manager failed to start.

Solution: Removing all nvidia related packages and reinstalling with mhwd with the new config: video-nvidia.

I can rollback to pre “BREAKAGE EXPECTED” state if you provide a new update for Nvidia to test the update process.

You can just use an old ISO in a VM.

There will be a totally new testing update in 7 minutes

2 Likes

New testing update here