Bug 1185542 - 32 bit Tumbleweed failing to boot with the 5.12.0-1-pae kernel must revert to 5.11.16-1-pae kernel.
32 bit Tumbleweed failing to boot with the 5.12.0-1-pae kernel must revert to...
Status: RESOLVED NORESPONSE
Classification: openSUSE
Product: openSUSE Tumbleweed
Classification: openSUSE
Component: Kernel
Current
32bit openSUSE Tumbleweed
: P5 - None : Normal (vote)
: ---
Assigned To: openSUSE Kernel Bugs
E-mail List
:
Depends on:
Blocks:
  Show dependency treegraph
 
Reported: 2021-05-03 01:45 UTC by Paul Graff
Modified: 2023-01-20 12:42 UTC (History)
4 users (show)

See Also:
Found By: ---
Services Priority:
Business Priority:
Blocker: ---
Marketing QA Status: ---
IT Deployment: ---
tiwai: needinfo? (pj.opensuse)


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description Paul Graff 2021-05-03 01:45:31 UTC
Hello to you all. I just obtained the latest openSUSE TW offline image (openSUSE-Tumbleweed-DVD-i586-Snapshot20210429-Media.iso) from the following: -> https://get.opensuse.org/tumbleweed

I am certain that the Compaq-nc6400 laptop is 32 bit not 64 bit.

Issue is after selecting "installation" within live bootmenu. The live system information loads and displays information lines up until pausing at "0.957026 trying to unpack rootfs image as initramfs...".

I previously have successfully installed a live 32 bit TW image (openSUSE-Tumbleweed-DVD-i586-Snapshot20210418-Media.iso) on the computer but the latest kernel 5.12.0-1-pae fails boot into splash screen and freezes the computer. I revert to kernel 5.11.16-1-pae then all boots to splash screen just fine.

This above is why I am attempting to install this latest snapshot.

What if any options do I have going forward with this?

Thank you
Comment 1 Felix Miata 2021-05-03 08:11:24 UTC
(In reply to Paul Graff from comment #0)
...
> What if any options do I have going forward with this?

I have 10+ TW installations on 32-bit Pentium IV PCs. All support a maximum of 2GB RAM, so I run all on kernel-default. I just upgraded one to TW20210430 and kernel 5.12.0. All seems normal. Maybe your approach to take is kernel-default. How much RAM do you have?
Comment 2 Paul Graff 2021-05-03 08:26:07 UTC
I believe I am using kernel defaults (I am not certain though). I have 4 gigabytes of DDR2 installed in this laptop since it's 32 bit.

Thanks
Comment 3 Paul Graff 2021-05-03 11:31:32 UTC
I would like to post a link to the forum question that initially led me to report this bug. https://forums.opensuse.org/showthread.php/553483-32-bit-TW-failing-to-live-boot-with-latest-snapshot

Thanks for your help.
Comment 4 Takashi Iwai 2021-05-03 15:10:18 UTC
Could you try to install kernel-default-5.12 and check whether it boots or not, as asked in comment 1?
Comment 5 Paul Graff 2021-05-03 23:21:16 UTC
Thank you for your response. I have installed kernel 5.12.1-1.g9bb8944-default (I hope this is correct in what you are referring to as kernel-default-5.12). I powercycled the computer and selected kernel-5.12.1-1.g9bb8944-default option. The system stalls at "Loading initial ramdisk ...".

I would like to add that prior to installing kernel 5.12.1-1.g9bb8944-default earlier, I performed a "zypper dup" which installed along with packages the kernel 5.1.2.0-2-pae (Working kernel 5.11... is no longer shown on the boot menu). 

Selecting kernel 5.12.0.2-pae does give some reporting errors as follows: 
System stalls and was reporting something like "CPU 0 stalled...." but not anymore just stops stuck at "ret_from_fork0x1c/0x20" or "Kernel Offset:disabled". Finally the system powercycles. 

Thanks
Comment 6 Paul Graff 2021-05-03 23:51:39 UTC
Additional report of the following with kernel-default once again:  watchdog: BUG: soft lockup - CPU#0 stuck for 22s! [kworker/0:3:82]

It takes alot of powercycles and waiting in order for this message to appear.

I really hope that I can use the computer again with the new kernel.

Thank you
Comment 7 Takashi Iwai 2021-05-04 08:33:29 UTC
OK, thanks, then it's a generic problem of 32bit Intel kernel.
Feel free to uninstall kernel-default again.

Now, we need more details about the crash.  Any chance that you have the relevant kernel Oops or panic in the saved kernel logs?  Or if not, try to boot without quiet boot option, and check whether you get the stack traces on screen, and take a picture.  That's often helpful as a start.
Comment 8 Paul Graff 2021-05-04 20:46:47 UTC
Thank you for your response. I would like to send you 3 weblinks that will direct to pictures I have taken of the "stack trace".

1. https://ibb.co/5Mtz9fL
2. https://ibb.co/S3g41VV
3. https://ibb.co/PwkND7F

I am working on getting kernel logs for you also if needed after this point. I hope that the above photos will in some way assist.

Please let me know on this when you can. I am not experienced in reporting bugs.

Thanks
Comment 9 Paul Graff 2021-05-05 02:12:42 UTC
I apologize for posting again but I found that using a camera would possibly be my only option. I would like to add 2 more photos of the "stack traces".

1. https://ibb.co/vZsM8Kq
2. https://ibb.co/djh9cwD

I do hope this helps to some degree.

-Regards
Comment 10 Jiri Slaby 2021-05-05 06:33:05 UTC
(In reply to Paul Graff from comment #9)
> 1. https://ibb.co/vZsM8Kq

smp_call_function_single+0xea -- what kernel flavor+version exactly is this with?
Comment 11 Paul Graff 2021-05-05 06:48:47 UTC
This version: -> 5.12.0-2-pae

This is the kernel version. I am not certain the meaning of kernel flavor exactly. I will report additional information if needed.

-Regards
Comment 12 Jiri Slaby 2021-05-05 06:53:13 UTC
(In reply to Jiri Slaby from comment #10)
> smp_call_function_single+0xea -- what kernel flavor+version exactly is this
> with?

Nevermind, it's likely 5.12.0-1. And the stack trace boils up to toggle_allocation_gate. That is kfence and kfence is new in 5.12.

Does adding kfence.sample_interval=0 to kernel parameter workaround the issue?
Comment 13 Jiri Slaby 2021-05-05 06:53:37 UTC
(In reply to Paul Graff from comment #11)
> This version: -> 5.12.0-2-pae
> 
> This is the kernel version. I am not certain the meaning of kernel flavor
> exactly.

Flavor is "pae" above :).
Comment 14 Paul Graff 2021-05-05 07:30:15 UTC
Does not appear to work the following message is displayed.

[     0.486287] x86/cpu: VMX (outside TXT) disabled by BIOS
[80.597573] watchdog: BUG soft lockup - CPU#0 stuck for 23s! [kworker/u44:0:8]
then very fast scrolling lines and halt as same as before -- 
[  155.213574] Kernel Offset: disabled

System then automatically reboots after a certain period of time.

I will double check adding kfence.sample_interval=0 to kernel parameter again but quite certain no joy.
Comment 15 Jiri Slaby 2023-01-20 12:42:20 UTC
This is a very old issue, and moreover on 32bit which doesn't drag much attention. If it still happen, please reopen.