Arch Linux

gnufied

Member

Registered: 2015-04-15

Posts: 13

Delayed Hibernate sometimes fails on Thinkpad T450s

I am following instructions here for setting up delayed hibernate (https://wiki.archlinux.org/index.php/Po … rvice_file) and while it mostly works but sometimes the system won't resume correclty.

I have tried to debug this and I found some differences in messages logged into log files when system hibernates.

************* Logs when system won't resume properly ****************************
Sep 03 19:03:56 hyperion kernel: PM: resume of devices complete after 775.989 msecs
Sep 03 19:03:56 hyperion kernel: PM: Finishing wakeup.
Sep 03 19:03:56 hyperion systemd[1]: Time has been changed
Sep 03 19:03:56 hyperion systemd[756]: Time has been changed
Sep 03 19:03:56 hyperion kernel: Restarting tasks ... done.
Sep 03 19:03:56 hyperion systemd-sleep[8462]: System resumed.
Sep 03 19:03:56 hyperion systemd[1]: Started Suspend.
Sep 03 19:03:56 hyperion systemd[1]: sleep.target: Unit not needed anymore. Stopping.
Sep 03 19:03:56 hyperion systemd[1]: Stopped target Sleep.
Sep 03 19:03:56 hyperion systemd[1]: suspend-to-hibernate.service: Unit not needed anymore. Stopping.
Sep 03 19:03:56 hyperion systemd[1]: tlp-sleep.service: Unit not needed anymore. Stopping.
Sep 03 19:03:56 hyperion systemd[1]: Stopping TLP suspend/resume...
Sep 03 19:03:56 hyperion systemd[1]: Stopping Delayed hibernation trigger...
Sep 03 19:03:56 hyperion sh[8491]: hibernate triggered
Sep 03 19:03:56 hyperion pulseaudio[891]: [pulseaudio] sink-input.c: Failed to create sink input: sink is
Sep 03 19:03:56 hyperion systemd[1]: Starting TLP suspend/resume...
Sep 03 19:03:56 hyperion systemd[1]: Started TLP suspend/resume.
Sep 03 19:03:56 hyperion systemd[1]: Reached target Sleep.
Sep 03 19:03:56 hyperion systemd[1]: Starting Hibernate...
Sep 03 19:03:56 hyperion systemd-sleep[8693]: Suspending system..


***************** Logs when system resumes correctly *************************************

Sep 04 07:58:16 hyperion kernel: PM: resume of devices complete after 777.057 msecs
Sep 04 07:58:16 hyperion kernel: PM: Finishing wakeup.
Sep 04 07:58:16 hyperion systemd[1]: Time has been changed
Sep 04 07:58:16 hyperion systemd[568]: Time has been changed
Sep 04 07:58:16 hyperion systemd[1028]: Time has been changed
Sep 04 07:58:16 hyperion kernel: Restarting tasks ... done.
Sep 04 07:58:16 hyperion kernel: psmouse serio1: synaptics: queried min coordinates: x [1266..], y [1096..
Sep 04 07:58:16 hyperion systemd-sleep[4523]: System resumed.
Sep 04 07:58:16 hyperion systemd[1]: Started Suspend.
Sep 04 07:58:16 hyperion systemd[1]: sleep.target: Unit not needed anymore. Stopping.
Sep 04 07:58:16 hyperion systemd[1]: Stopped target Sleep.
Sep 04 07:58:16 hyperion systemd[1]: tlp-sleep.service: Unit not needed anymore. Stopping.
Sep 04 07:58:16 hyperion systemd[1]: suspend-to-hibernate.service: Unit not needed anymore. Stopping.
Sep 04 07:58:16 hyperion systemd[1]: Stopping Delayed hibernation trigger...
Sep 04 07:58:16 hyperion systemd[1]: Stopping TLP suspend/resume...
Sep 04 07:58:16 hyperion sh[4535]: hibernate triggered
Sep 04 07:58:16 hyperion systemd[1]: Starting TLP suspend/resume...
Sep 04 07:58:16 hyperion systemd[1]: Started TLP suspend/resume.
Sep 04 07:58:16 hyperion systemd[1]: Reached target Sleep.
Sep 04 07:58:16 hyperion systemd[1]: Starting Hibernate...
Sep 04 07:58:16 hyperion systemd-sleep[4692]: Suspending system...
Sep 04 07:58:16 hyperion kernel: PM: Hibernation mode set to 'platform'
Sep 04 08:00:32 hyperion kernel: PM: Syncing filesystems ... done.
Sep 04 08:00:32 hyperion kernel: Freezing user space processes ... (elapsed 0.001 seconds) done.

Notice how when hibernate works correctly, the messages logged say:

Sep 04 07:58:16 hyperion kernel: PM: Hibernation mode set to 'platform'
Sep 04 08:00:32 hyperion kernel: PM: Syncing filesystems ... done.

Also, I am using swap as hibernate partition. I have resume hook correctly in mkinitcpio.conf file and resume line in Kernel boot parameter. However, my Swap parition size is only 8GB whereas my system has 12GB of RAM.

Last edited by gnufied (2015-09-06 15:04:24)

gnufied

Member

Registered: 2015-04-15

Posts: 13

Re: Delayed Hibernate sometimes fails on Thinkpad T450s

After I changed swap partiton size to be more than RAM size, this problem seems to have gone away. The arch wiki indicates SWAP partition can be smaller than RAM size but it appears that, a setup like that causes some instability.

Last edited by gnufied (2015-09-05 11:38:41)

gnufied

Member

Registered: 2015-04-15

Posts: 13

Re: Delayed Hibernate sometimes fails on Thinkpad T450s

So this is not really fixed by increasing SWAP parition size. Sure, 9 out of 10 times, hibernate and resume works properly but it still can fail 1 time from what I have seen. Whenever that happens, the hibernate logs don't seem to have the line "Syncing file system.. done" in it. So I presume whenever hibernate/resume fails it was because the syncing filesystems did not go through properly.

etoixpi

Member

Registered: 2015-04-16

Posts: 8

Re: Delayed Hibernate sometimes fails on Thinkpad T450s

******UPDATE******
This was found to be a kernel bug, but the developers need more information. If you are experiencing the same issue please post to: https://bugzilla.kernel.org/show_bug.cgi?id=104771

drrossum

Member

From: Chicago

Registered: 2009-02-24

Posts: 83

Re: Delayed Hibernate sometimes fails on Thinkpad T450s

For me the resume seems to fail more often when the power adapter is connected. Do you guys also see that correlation?

---

x-selection paste with keyboard in addition to center mouse button

ddurdle

Member

Registered: 2016-06-04

Posts: 1

Re: Delayed Hibernate sometimes fails on Thinkpad T450s

The issue is not limited to just the T450 -- all modern Thinkpads, it would appear.  Been suffering this issue with Carbon 2nd generation.  Been trying everything.  Seems very random when it fails.  Been keeping a log of what is running, power settings, whether AC or running off battery, etc, when I hibernate, so I can track when it fails.  For longest time, I thought if you startup on battery, it would likely fail.  But been seeing it fail when starting up on AC.  Only time I have not seen it fail is when I've left the PC plugged into the AC without ever removing it at any time (whether system is off or on).

I just got a Carbon 4th generation this week -- same issue of course.