FS#48596 - [linux] Kernel 4.5: AMD E-2000 CPU hang at boot

Attached to Project: Arch Linux
Opened by AK (Andreaskem) - Tuesday, 15 March 2016, 17:07 GMT
Last edited by Tobias Powalowski (tpowa) - Sunday, 26 June 2016, 05:54 GMT
Task Type Bug Report
Category Packages: Testing
Status Closed
Assigned To Tobias Powalowski (tpowa)
Thomas Bächler (brain0)
Architecture All
Severity Low
Priority Normal
Reported Version
Due in Version Undecided
Due Date Undecided
Percent Complete 100%
Votes 2
Private No

Details

I just updated to kernel 4.5 in [testing] and twice in a row, the boot failed with a CPU hang reported at different times. At some point the NMI watchdog kicked in. I had to hard reboot the system. I currently don't have the time to debug this and just want to leave a warning here.

CPU: AMD E-2000 with integrated Radeon HD 7340. System: Lenovo ThinkPad Edge E135
This task depends upon

Closed by  Tobias Powalowski (tpowa)
Sunday, 26 June 2016, 05:54 GMT
Reason for closing:  Upstream
Comment by AK (Andreaskem) - Tuesday, 15 March 2016, 19:01 GMT
And for some reason I am unable to downgrade the kernel. No matter what I try, it always somehow tries to boot version 4.5. That is really weird.

edit: I even tried installing linux-lts which went fine. But when trying to boot it, a message about missing modules from 4.5 is printed, the system does not find the root disk by its usual UUID and I get dropped to a shell. This shell is useless and I managed to hang the system several times with some commands. Mkinitcpio seemed to do its job with the LTS kernel. I have no idea what's going on. I never had any problems downgrading a kernel before.

edit2: Okay, after I stopped being an idiot, I managed to get linux-lts working after all.
Comment by KriK (krik) - Wednesday, 16 March 2016, 06:24 GMT
Same problem, but with i7-3537U CPU
Comment by AK (Andreaskem) - Saturday, 16 April 2016, 10:27 GMT
A few minutes ago I gave kernel 4.5.1 a shot and the issue was still present. I decided to spend a few minutes debugging the early boot process. Turning on debug output on the kernel command line led to an endless mess of error messages that went by too fast to decipher. I thought I spotted something about ACPI in there so I added acpi_osi="!Windows 2012" which I used to have turned on for backlight issues. After that (or by pure chance) the endless stream of messages became decipherable.

Long story short, my issue is the following:
https://bugzilla.kernel.org/show_bug.cgi?id=114201

It can be "fixed" by blacklisting sp5100_tco. This allows my system to boot without issues.
Comment by Finke Lamein (Quattro) - Thursday, 19 May 2016, 21:01 GMT
4.5.4 is stable (and outdated now).

4.6-1 is now in testing, is the issue still present?

No issues on my core i5 6600 by the way.
Comment by AK (Andreaskem) - Tuesday, 24 May 2016, 15:39 GMT
No, this issue is not fixed. The kernel bug report has a patch available (I did not test it) but it has not landed in a release yet, apparently. Just to confirm, I tried to boot 4.6 without blacklisting sp5100_tco and the boot process did not finish.

Loading...