today's update problem

Hi,
Not sure if this might be noted here or not, but new leap 15.1 update released today prevent my sever to startup.
It hangs after boot menu disappears.
As I use btrfs I use snapper rollback and works fine.
I use AMD threadripper 2950 and WX7100 radeonpro on gigabyte motherboard.

@neocurve:

Here, also with an AMD CPU, chipset and Graphics – the following patches which have occurred since Tuesday this week, haven’t caused any issues at reboot:

  • 2020-01-14 15:47:12|install|kernel-firmware|20191118-lp151.2.9.1|
  • 2020-01-14 15:47:16|install|ucode-amd|20191118-lp151.2.9.1|

The optional patch “openSUSE-2020-52” also hasn’t caused any issues at reboot:

  • 2020-01-16 15:02:51|install|openslp|2.0.0-lp151.7.6.1|

I installed it with “zypper update” – could also have used “zypper install patch:openSUSE-2020-52”

[HR][/HR]Can you log into a VT session on the affected server?

  • If so, can you please take a look at the systemd journal for the boot which failed?

“journalctl --boot=[ID]±offset]” or “journalctl -b [ID]±offset]”
Use “journalctl --list-boots” to work out the ID or the offset of the boot which failed.

Here is the list of my zypper update :

The following NEW package is going to be installed:
  libvo-amrwbenc0

The following 52 packages are going to be upgraded:
  cpp7 gcc7 gcc7-ada gcc7-c++ gcc7-fortran gcc7-info kernel-firmware libada7
  libasan4 libavcodec58 libavdevice58 libavfilter7 libavformat58 libavresample4
  libavutil56 libcilkrts5 libgcj-gcc6 libgfortran4 libGraphicsMagick3-config
  libGraphicsMagick++-Q16-12 libGraphicsMagick-Q16-3 libobjc4 libopenssl1_1
  libopenssl1_1-32bit libpostproc55 libstdc++6-devel-gcc7 libstorage-ng1
  libstorage-ng-lang libstorage-ng-ruby libswresample3 libswscale5 libubsan0
  libvirglrenderer0 libvlc5 libvlccore9 libx265-179 libyui-qt-pkg9 openslp
  openssl-1_1 python3-acme python3-certbot python3-certbot-apache
  python3-certbot-dns-rfc2136 python3-certbot-dns-route53 python3-certbot-nginx
  ucode-amd vlc vlc-codec-gstreamer vlc-lang vlc-noX vlc-qt vlc-vdpau

The following 2 packages require a system reboot:
  kernel-firmware libopenssl1_1

after update, I cannot even go to VT session. :frowning:
This is the first time happened with my Leap!!

Looking at that update list, most of them should not cause the problem that you describe. So my best guess is that it is the “ucode-amd” that is causing your problems.

If you have enough access, you could try going back to the previous version. Or use your btrfs rollback to get a working system, then lock “ucode-amd” and try updating again with that package locked (to prevent update).

@neocurve:

The 3 most recent Leap 15.1 “ucode-amd” updates I have here are:


2019-07-02 18:38:12|install|ucode-amd|20190312-lp151.2.3.1|
2019-07-21 15:41:57|install|ucode-amd|20190618-lp151.2.6.1|
2020-01-14 15:47:16|install|ucode-amd|20191118-lp151.2.9.1|

In the change history for “ucode-amd” there’s the following:

  • Drop the previous vega10_sos.bin revert, as already addressed in
    the new firmware
  • Revert amdgpu/vega10_sos.bin to the previous version for fixing
    the hang up at boot time (bsc#1143331)

Whether or not, your AMD CPU and/or graphics card are suffering something similar to the Vega 10 issue, is a moot point …

  • Please raise an openSUSE Bug Report against the “ucode-amd” package detailing what we’ve been discussing here.

ucode-amd might not be the only one issue for this update.
I have locked the package and updated leap but got the same result.
I raise a bug to see if can go for smooth update.

Thanks

Are you using the amdgpu driver or, the radeon driver?

  • RadeonTM Pro WX 7100 is a “Polaris” – Arctic Islands – GCN4.x card.

Recommended is the amdgpu driver, probably without any kernel command line parameter set but, “amdgpu.ngg=1” may, possibly, be needed.

VGA compatible controller: Advanced Micro Devices, Inc. [AMD/ATI] Ellesmere [Radeon Pro WX 7100]
    Subsystem: Advanced Micro Devices, Inc. [AMD/ATI] Device 0b0d
    Kernel driver in use: amdgpu
    Kernel modules: amdgpu

So, it looks fine I think.
Not sure how I can see or change “amdgpu.ngg=1”

Easiest way is with

Yast –> System –> Boot Loader

then click on the “Kernel Parameters” tab.

Thank you very much for your advice.
I needed Radeon pro driver for my GPU and 15.1 is not supported at the moment.
I just deleted my Leap and went for another supported distro.
Thank you very much BTW for the help.

Moving to the previous version of kernel-firmware might probably solve the problem.

At the moment there are four different kernel-firmware packages available in the openSUSE Leap 15.1 repositories

kernel-firmware | 20191118-lp151.2.9.1 | Main Update Repository
kernel-firmware | 20190618-lp151.2.6.1 | Main Update Repository
kernel-firmware | 20190312-lp151.2.3.1 | Main Update Repository
kernel-firmware | 20190118-lp151.1.10 | Main Repository

On my laptop i have openSUSE Tumbleweed as primary and openSUSE Leap 15.1 as fallback system. I keep the latter updated on a regular basis.

A few days ago openSUSE Leap 15.1 started to die very early in the boot process with a kernel panic (somewhere in the iwlwifi driver).

Last night i found out that reverting kernel-firmware 20191118-lp151.2.9.1 to kernel-firmware 20190618-lp151.2.6.1 removed the problem.

Regards

susejunky

I wish I could see your message earlier :frowning: Leap was on my main machine and could not use it with that problem.
I liked Leap and was great distro and I might back again if I don’t feel comfortable with new distro, so just to learn more to solve such issues in future, can I do that kernel transition by YaST to deactivate it or I need to completely delete that kernel firmware ?

Do you mean how to go back to an earlier version of a package as @susjunky did?

Start YaST > Software > Software Management. In the Search View search for the pcakage. At right select the package, then below select the Version tab. There you see the versions available as @susejunky showed them above. Choose the one you like and install. Then probably the best is to go back to that package and block it until you are sure a newer version is published that does have a patch for your problem.

Basically it works as described by Henk van Velden in his post.

However if you are hit by the problem like i was and your system won’t even boot anymore then a few additional steps are needed:

  1. Boot from a different environment (like rescue system from the openSUSE installation DVD, openSUSE Live media, etc.)
  2. chroot
    into the not-working system 1. use YaST
    to revert to the lower version of the kernel-firmware 1. reboot (your “normal” system)
  3. use** YaST**
    to lock the kernel-firmware-package so that it is no longer updated

If you need any support just ask here.

Regards

susejunky

May I please note that nothing will happen without a bug report? The packagers and devs hardly use these forums.

Sorry, but my experience is that even with a bug report nothing will happen.

Considering the fact that Tumbleweed already got a newer Version of kernel-firmware (20200114-1.1 which works flawless) the problem might be gone before someone has a look at my bug report.

Sorry again, but NO i am not willing to waste my time on bugzilla reports any longer.

Regards

susejunky

Just for the sake of completeness:

https://forums.opensuse.org/showthread.php/538814-kernel-firmware-20191118-lp151-2-9-1

https://bugzilla.opensuse.org/show_bug.cgi?id=1161243

Regards

susejunky

Thanks to all of you for the great support :slight_smile: