0

Ubuntu 22.10 doesn't boot anymore. Upon boot I get the following messages:

[   1.829574] usb 1-3.4: device descriptor read/64, error -32
/dev/sda6: clean, 271463/9601024 files, 15565532/38384128 blocks
[   6.441227] nvidia-gpu 0000:01:00.3: i2c timeout error e0000000
[   6.442378] ucsi_ccg 0-0008: i2c_transfer failed -110
[   6.443371] ucsi_ccg 0-0008: ucsi_ccg_init failed - -110
[ 256.957586] snd_hda_intel 0000:00:1b.0: Too many BDL entries: buffer=4194300, period=135300
[ 319.513256] snd_hda_intel 0000:00:1b.0: Too many BDL entries: buffer=4194300, period=135300

The screen occasionally blinks with more messages but it's too fast to read, and comes back to these messages with an additional snd_hda_intel "Too many BDL entries" error.

The kernel must have been updated from 5.19.0-31 to 5.19.0-35 because I have these two entries in GRUB's advanced options.

The system boots normally with the older kernel version.

I had the same problem previously after a kernel update and I had to reinstall the OS. Let's hope this won't be necessary this time.

Sorry, the question is very vague. please tell me if there is any way I can gather more info.

Cutter
  • 11
  • So, from what I understand, your kernel got updated to 5.19.0-35 and it won't boot unless in recovery mode. But there's also an entry for an older 5.19.0-31 kernel — does this one boot normally? – Hi-Angel Mar 08 '23 at 13:21
  • Yes 5.19.0-31 boots without a problem. – Cutter Mar 08 '23 at 13:24
  • I see. So, in this case your further actions depend on what you want. You can remove the newer package (I think that'd be sudo apt purge linux-image-5.19.0-35-generic, check that it doesn't remove any other kernel version) and avoid updating the kernel for now. You might also want to create a bugreport about this situation on launchpad, to make sure Ubuntu kernel maintainers are aware of this problem, so they'd find a fix. Since the 35 number signifies custom patches, most likely some of the patches maintainers applied to the kernel screwed something up. – Hi-Angel Mar 08 '23 at 13:32
  • @Hi-Angel when I enter the command you gave, apt warns me that it is about to remove the following packages: linux-generic-hwe-22.04* linux-image-5.19.0-35-generic* linux-image-generic-hwe-22.04*, should I do this ? – Cutter Mar 09 '23 at 20:25
  • @karel how can I pin the previous kernel in grub ? Especially since it is hidden in the advanced options. – Cutter Mar 09 '23 at 20:26
  • @Cutter hm, no, you don't want to remove anything except the specific kernel version. Try replacing purge with remove in my command. Although, I find it odd that it suggests removing the meta-packages, given there is another kernel version that belongs to them… Just for the safe case, before removing, please also make sure there is output for this command dpkg -l | grep linux-image-5.19.0-31 – Hi-Angel Mar 10 '23 at 07:21
  • I would like grub to remove the new kernel entry and have the "Ubuntu" entry on the main menu point to the previous kernel. How can I do that ? @Hi-Angel is there a boot log that I could provide in my bug report ? – Cutter Mar 12 '23 at 11:48
  • @Cutter assuming that / gets mounted correctly, and journal gets flushed before you turn off the computer, you will find the kernel logs from unsuccessful boots at /var/log/ dir, files with prefix dmesg or kern. You can make sure you found the correct log by looking up the error messages about BDL entries that you know should show up. And regarding you first question about grub entries — simply removing the kernel version that is known to be broken will make default Grub entry point to the older kernel. – Hi-Angel Mar 12 '23 at 12:57

2 Answers2

0

Just hit the same issue here (with 5.19.0.35), purging the new kernel did the trick, before restarting just make sure the old version is still installed.

Thanks for the help.

0

I still have this problem.

I just had to get back to 5.19.0-32.

35 is not working.

Seems to be something also with nvidia-driver-525.