Created attachment 284477 [details] lshw output, syslog (debian 9, debian 10) I noticed this issue at clean Debian Buster instalation (kernel 4.19). Reboot works. On kernels older than 4.18.0 i dont have this issue. I used different distros 64-bit and all have same behavior. I figured that "systemctl poweroff" working fine till kernel 4.18.0, then not. I made clean Debian 9 64-bit minimal install (no desktop). First login as root and "systemctl poweroff" power off my notebook. I did same with Debian 10 64-bit, notebook "freeze" at "systemctl poweroff": last line on monitor was "reboot: Power down." I power off laptop with helding power off button on notebook for 3 sec. Windows10 - no power off issue. I use UEFI mode for all instalations. PS: no problems with reboot
please confirm if the problem also exist in the latest upstream kernel.
(In reply to Zhang Rui from comment #1) > please confirm if the problem also exist in the latest upstream kernel. Problem still exist: Debian Release 10 (buster) 64-bit Kernel Linux 4.19.0-5-amd64 x86_64
this seems like a kernel regression. can you run git bisect to find out which commit introduces the problem?
I ran git bisect (https://wiki.debian.org/DebianKernel/GitBisect): ------------------------------------------------------------------ b7ccc7f8c643de0b71cd2da3245d0c27038a8dd7 is the first bad commit commit b7ccc7f8c643de0b71cd2da3245d0c27038a8dd7 Author: Matthew Wilcox <mawilcox@microsoft.com> Date: Thu Jun 7 17:08:46 2018 -0700 mm: move lru union within struct page Since the LRU is two words, this does not affect the double-word alignment of SLUB's freelist. Link: http://lkml.kernel.org/r/20180518194519.3820-10-willy@infradead.org Signed-off-by: Matthew Wilcox <mawilcox@microsoft.com> Acked-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Christoph Lameter <cl@linux.com> Cc: Dave Hansen <dave.hansen@linux.intel.com> Cc: Jérôme Glisse <jglisse@redhat.com> Cc: Lai Jiangshan <jiangshanlai@gmail.com> Cc: Martin Schwidefsky <schwidefsky@de.ibm.com> Cc: Pekka Enberg <penberg@kernel.org> Cc: Randy Dunlap <rdunlap@infradead.org> Cc: Andrey Ryabinin <aryabinin@virtuozzo.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> :040000 040000 81de2d449778b523de1aee6382ab2997e919eb25 b53493e18b633843d4b97fb74e7fd3e0d508c7cf M include :040000 040000 9a9d1458f0308c6d337816509e4cc63b5475db5f a2fe407b7973fc426427c478be9e16d0a794b62b M mm
so please confirm is the problem exists with this commit, and does not exist with the commit previous to this one.
Yes, i confirm that the problem exists with this commit, and does not exist with the commit previous to this one.
This makes no sense. I can see no way in which this commit can cause that problem.
I have an old computer (core 2 duo, no laptop) too. I install Debian 10 on it and i have no issue. I tried differrent linux distros on Medion Erazer x7830 laptop (Manjaro, Fedora, Debian, Ubuntu, Solus) with kernel newer than 4.18 and power off is no work. With older distros (with kernel older than 4.17) power off work fine. I got this coomit with git-bisect. I dont have knowledge, but i dont see connection with this commit and my problem. I have no idea, maybe is reason systemd - i will wait on new Devuan distro (it not have systemd). Now i use Debian 9 - with no issue. It has support yet. For Debian 10 (or other newer linux distro) is rebooting into Debian 9 for power off the computer is some solution. Thank you for your help.
(In reply to Zhang Rui from comment #5) > so please confirm is the problem exists with this commit, and does not exist > with the commit previous to this one. I got the answer that commit i found with git-bisect had no sense with my problem. I try pclinuxos distro too (it has no systemd), the problem remaining. I now use hibernate instead power off, and this works. I have no idea what is the correct solution or cause for this problem began with kernel 4.18. The last BIOS update for Medion Erazer x7830 is from 2015. Thank you for your help.
I carefully repeat git-bisect v4.17..v418rc1 and i got same commit (comment 4). This commit change file include/linux/mm_types.h. I got kernels 4.17 and 4.18. I had overwrite mm_types.h from 4.18 to 4.17 and build modified 4.17 kernel. It not poweroff (i expected this). Original 4.17 kernel has no poweroff issue. Additional info: suspend works in Buster, in Stretch not. poweroff works in Stretch, in Buster not (this reported bug). workaround: in Buster use hibernate instead poweroff - it take 2-3 x more time.
fix last comment: suspend works in Stretch too.
does the problem still exists in latest upstream kernel? This problem looks tricky and I have no idea what could cause the problem. BTW, a simple way to confirm your bisect result is to 1. git checkout b7ccc7f8c643de0b71cd2da3245d0c27038a8dd7 and confirm if poweroff is broken 2. git checkout b7ccc7f8c643de0b71cd2da3245d0c27038a8dd7~ and confirm if poweroff is working