Bug 1172892

Summary: Boot hangs on initrd: ucode-intel-20200609-1.1
Product: [openSUSE] openSUSE Tumbleweed Reporter: Tuukka Pasanen <tuukka.pasanen>
Component: BasesystemAssignee: E-mail List <screening-team-bugs>
Status: RESOLVED DUPLICATE QA Contact: E-mail List <qa-bugs>
Severity: Major    
Priority: P5 - None CC: rickscafe.casablanca
Version: Current   
Target Milestone: ---   
Hardware: x86-64   
OS: SUSE Other   
Whiteboard:
Found By: --- Services Priority:
Business Priority: Blocker: ---
Marketing QA Status: --- IT Deployment: ---

Description Tuukka Pasanen 2020-06-14 09:18:28 UTC
After zypper dup newest Tumbleweed and rebooting systems most of the time hangs on loading initrd and I haven't found rd.break level to debug it. It boots with older 5.5.11 kernel sometime (every tenth or something reboot) if you add parameter debug to slow it down and then it breaks to dracut debug as expected. It could work with 5.6.x series kernel but I haven't got it booting event with several attempts.
If there is some way to get early debug log please note me to get into bottom of this problem
Comment 1 Mark Draheim 2020-06-14 18:56:05 UTC
same here. For me it is with kernels kernel-default-5.6.12-1.3.x86_64,
kernel-default-5.6.14-1.6.x86_64. The older one seems to have a better chance to actually boot but most of the time I get this "Loading initial ramdisk" and then it just sits there. I had a look at initrd contents but they seem to be pretty identical but for the kernel modules. It started after recent updates, here done on 2020-06-13. System is encrypted root with unencrypted /boot, all on btrfs.
Comment 2 Mark Draheim 2020-06-17 16:46:54 UTC
I think it is related to intel microcode loading. When I disable that via grub linux switch dis_ucode_ldr, all kernels boot fine.

For reference, this one year old bug

https://github.com/intel/Intel-Linux-Processor-Microcode-Data-Files/issues/1

Incidentially, my laptop is an ASUS Zenbook UX303UB with this CPU

vendor_id       : GenuineIntel
cpu family      : 6
model           : 78
model name      : Intel(R) Core(TM) i5-6200U CPU @ 2.30GHz
stepping        : 3
microcode       : 0x74

and I do have bootloader set to mitigations=off
Comment 3 Mark Draheim 2020-06-17 17:13:07 UTC
Asus released a bios update for my laptop last year. I was not aware that Asus still maintains this 2016 model. That bios comes with microcode version 0xba which is still less than the 0xdc that grub would load.

I still need the dis_ucode_ldr param, though.
Comment 4 Tuukka Pasanen 2020-06-18 07:11:28 UTC
(In reply to Mark Draheim from comment #3)
> Asus released a bios update for my laptop last year. I was not aware that
> Asus still maintains this 2016 model. That bios comes with microcode version
> 0xba which is still less than the 0xdc that grub would load.
> 
> I still need the dis_ucode_ldr param, though.

It seems that this Asus UX**** problem as I also have same old Asus with Intel M-processor and dis_ucode_ldr fixes issue
Comment 5 Mark Draheim 2020-06-18 07:24:37 UTC
So, there is a workaround but it would still be great to have a real fix. Asus laptops aren't exactly rare ;-)

Strange thing is that all three of my installed kernels apperently have the same intel ucode blob in front of their initrds. But only the oldest 5.6.12 boots (sometimes) whereas the newer ones fail consistently.

Anyway, can I take the liberty and edit this bug's subject line? Will do
Comment 6 Tuukka Pasanen 2020-06-18 07:53:04 UTC
(In reply to Mark Draheim from comment #5)
> Anyway, can I take the liberty and edit this bug's subject line? Will do

Be my guest new topic is more precises.

I'm just wondering is this kernel problem. Got to test with vanilla flavor build without any (open)SUSE patching. If it's still hanging then it's something to-do with tooling or this microcode.
Comment 7 Tuukka Pasanen 2020-06-18 07:58:55 UTC
Just pasting my CPU info also. My microcode is higher than your if there is anything to todo with anything.

vendor_id	: GenuineIntel
cpu family	: 6
model		: 78
model name	: Intel(R) Core(TM) m3-6Y30 CPU @ 0.90GHz
stepping	: 3
microcode	: 0xba
Comment 8 Mark Draheim 2020-06-18 08:48:06 UTC
Right, maybe I can find the previous ucode-intel still on my home server which has a slower update cycle. I once found a TW mirror that kept old version but I lost the URL.

If boot also fails with the older blob, then it might be tooling after all.
Comment 9 Mark Draheim 2020-06-19 20:12:15 UTC
it is the ucode. And this ticket here is a dup of #1172856
Comment 10 Mark Draheim 2020-06-19 20:23:01 UTC
finally found the "mark as duplicate" button. Closing this one as all the action is in the other. And it seems a fix is on the way. Thanks

*** This bug has been marked as a duplicate of bug 1172856 ***