You are not logged in.

#1 2017-07-29 15:35:21

kdall99
Member
Registered: 2015-06-12
Posts: 17

Resume from hibernate fails with kernel 4.12.3

Hi,
since the upgrade to kernel 4.12.3 resuming from hibernate fails on my Asus eeePC 1005PE.
Downgrading the kernel to v. 4.9.11 (or using the kernel from linux-lts) resolves the problem.

With kernel 4.12.3, at first, the boot/resume procedure seems normal (lots of disk activity) but roughly at the time when kernels <4.12 switch to graphics mode and show the plasma lock screen, kernel 4.12.3 simply reboots. Removing "quiet" from the kernel command line doesn't give additional info about the problem. Setting acpi_osi doesn't help.

I couldn't find anything related in the Arch bug tracker. Nevertheless, is this a known problem? Is there a workaround?

System is a 32-bit Arch with LVM on LUKS and Plasma as a DE.

Regards

Offline

#2 2017-07-30 00:20:46

boehlke
Member
Registered: 2014-12-14
Posts: 5

Re: Resume from hibernate fails with kernel 4.12.3

I can confirm that this happened to me upon upgrading from 4.11.9 to 4.12.3.  The only entry from my log after I start hibernating and before I have to reboot is

 PM: Syncing filesystems ... 

This is what the log shows upon successful hibernation under linux-lts:

Jul 28 18:35:42 [kernel] [  104.397586] PM: Syncing filesystems ...
Jul 28 18:35:42 [kernel] [  104.438504] done.
Jul 28 18:35:42 [kernel] [  104.438511] Freezing user space processes ... (elapsed 0.001 seconds) done.
Jul 28 18:35:42 [kernel] [  104.440268] PM: Preallocating image memory... done (allocated 59958 pages)
Jul 28 18:35:42 [kernel] [  104.636184] PM: Allocated 239832 kbytes in 0.19 seconds (1262.27 MB/s)
Jul 28 18:35:42 [kernel] [  104.636186] Freezing remaining freezable tasks ... (elapsed 0.001 seconds) done.
Jul 28 18:35:42 [kernel] [  104.637844] Suspending console(s) (use no_console_suspend to debug)
Jul 28 18:35:42 [kernel] [  104.976418] PM: freeze of devices complete after 338.565 msecs
Jul 28 18:35:42 [kernel] [  105.000187] PM: late freeze of devices complete after 23.764 msecs
Jul 28 18:35:42 [kernel] [  105.001655] PM: noirq freeze of devices complete after 1.461 msecs
Jul 28 18:35:42 [kernel] [  105.002100] ACPI: Preparing to enter system sleep state S4
Jul 28 18:35:42 [kernel] [  105.004206] PM: Saving platform NVS memory
Jul 28 18:35:42 [kernel] [  105.004208] Disabling non-boot CPUs ...
Jul 28 18:35:42 [kernel] [  105.005556] smpboot: CPU 1 is now offline
Jul 28 18:35:42 [kernel] [  105.006155] PM: Creating hibernation image:
Jul 28 18:35:42 [kernel] [  105.014436] PM: Need to copy 58875 pages

Seems to fail pretty early in the process, although the disk activity seems normal.  I am also running a 32-bit system (Dell Latitude D420).

Offline

#3 2017-07-30 07:46:31

adesh
Member
Registered: 2016-10-05
Posts: 167

Re: Resume from hibernate fails with kernel 4.12.3

Offline

#4 2017-07-30 21:21:25

kdall99
Member
Registered: 2015-06-12
Posts: 17

Re: Resume from hibernate fails with kernel 4.12.3

adesh wrote:

Thanks for the tip! But unfortunately the 4.12.4 kernel I installed from testing doesn't solve the resume problem for me.
There is a difference, though: while 4.12.3, instead of showing the lock screen, simply rebooted, 4.12.4 hangs completely.

Offline

#5 2017-07-30 21:30:32

diegoviola
Member
Registered: 2010-12-16
Posts: 134

Re: Resume from hibernate fails with kernel 4.12.3

Deleted.

Last edited by diegoviola (2023-04-17 14:58:50)

Offline

#6 2017-07-30 22:56:35

diegoviola
Member
Registered: 2010-12-16
Posts: 134

Re: Resume from hibernate fails with kernel 4.12.3

Deleted.

Last edited by diegoviola (2023-04-17 14:58:57)

Offline

#7 2017-07-30 23:28:20

kdall99
Member
Registered: 2015-06-12
Posts: 17

Re: Resume from hibernate fails with kernel 4.12.3

diegoviola wrote:

Any chance you can use 'git bisect' to find the offending commit?

No. I don't really have an idea how to do that.

Offline

#8 2017-07-30 23:49:54

loqs
Member
Registered: 2014-03-06
Posts: 18,961

Re: Resume from hibernate fails with kernel 4.12.3

kdall99 wrote:
diegoviola wrote:

Any chance you can use 'git bisect' to find the offending commit?

No. I don't really have an idea how to do that.

https://bbs.archlinux.org/viewtopic.php … 5#p1700245
Although that PKGBUILD is targeted at bisecting 4.9 to 4.10 it should be readily adaptable to 4.11 to 4.12

Offline

Board footer

Powered by FluxBB