Suspend broken after upgrade to 2.6.38 on Lenovo L512

Micand · 2011-04-22 16:19:22

Hello!

Since upgrading to kernel26 2.6.38.3-1, suspend has been broken on my Lenovo L512 laptop. Though putting the system into a suspended state works as always, waking it up freezes the system. Upon resuming from suspend, the entire screen is black, with a single white underscore in the top-left corner. The system is entirely unresponsive in this state -- its network connection is not restored (and so it cannot be SSHed into), it does not respond to CTRL-ALT-DEL, and pressing the power button to trigger a clean shutdown via an ACPI event doesn't work. The contents of /var/log/pm-suspend.log do not indicate anything abnormal. The same issue is present in 2.6.39-rc4. Suspend works somewhat in 2.6.37 -- restoring from suspend works, but half the time it seems the hard disk isn't "woken up," so that though the system initially appears to function normally, as soon as any program hits the hard disk weird things start happening. The last kernel where suspend and resume worked reliably was 2.6.36.

Any assistance will be much appreciated. Thanks!

pyther · 2011-04-22 22:18:43

Sounds like a kernel bug. I would search for bugs upstream. If you can't find anything you could open a new bug (upstream). However, I would highly recommend doing a git bisect to figure out what commit broke the suspend functionality.

This is a good guide on how to use git bisect to determine the bad commit: http://www.reactivated.net/weblog/archi … l-patches/

Micand · 2011-04-29 18:59:06

Thank you for your reply, pyther. I figured it was a kernel bug, too, but now I'm not so sure. Surprisingly, in Ubuntu 11.04 (which uses kernel 2.6.38), suspend works fine on my laptop. Figuring that perhaps this was due to an Ubuntu-specific patch, I tried booting the 2.6.38-8-generic kernel that ships with Ubuntu 11.04 on my Arch installation, but found that it was subject to the same issue described above, where the system fails to resume from suspend. I also tried disabling laptop-mode-tools, but to no avail.

Clearly, the problem is related somewhat to the kernel, as in my Arch installation, suspend works fine in 2.6.36, is somewhat problematic in 2.6.37, and is completely broken in 2.6.38. Yet I don't think it's entirely due to the kernel, as suspend works fine in Ubuntu 11.04, but is broken when I boot my Arch installation with Ubuntu's kernel. Performing a bit bisect to determine the bad commit sounds like it might help, but given that suspend functionality seems to have gotten progressively worse for me since 2.6.36, I'm not sure how I would go about locating a specific commit that is the source of my issues. Is there anything else I might try?

secretrobotron · 2011-04-29 19:25:02

Until very recently, I had a lot of problems suspending with my W510. The ThinkWiki certainly helped: http://www.thinkwiki.org/wiki/Category:W510.

Unfortunately, the ThinkWiki for the L512 is comparatively blank (http://www.thinkwiki.org/wiki/Category:L512), but I suggest trying some of the workarounds like putting `xhci_hcd` in `SUSPEND_MODULES`, and putting ` acpi_sleep=sci_force_enable` on your kernel line in menu.lst (if you're using grub).

If you're really stuck, try this out: https://wiki.ubuntu.com/DebuggingKernelSuspend. You'll have to do the Ubuntu -> Arch translation yourself, but it's worth a shot.

Very frustrating, I know.

rocknrolf77 · 2011-05-02 08:39:48

Same problem here, exept Ubuntu does not work either. 26.37 was no problem for me. My machine is a Pavilon DV6-2057eo. Sad that a newer kernel breaks stuff. (If that's the problem that is)

Soumyadeep · 2011-05-07 10:53:31

Exact same prob, 2.6.37 worked, 2.6.38 doesnt, my machine is an AMD desktop.

lkraav · 2011-05-07 14:12:29

i'm with the same issue running zen-kernel. i've now identified that going from 2.6.38.2 -> .3 is where suspend stops working, and as of .5 it has not been recovered yet. proceeding to bisect.

http://lwn.net/Articles/438548/ here's the vanilla kernel changelog.

Last edited by lkraav (2011-05-07 14:13:25)

MadCat_X · 2011-05-07 15:43:53

I do not suffer from this problem so I can't test it, but could this possibly be related?

Rafael J. Wysocki (1):
      PCI/ACPI: Report ASPM support to BIOS if not disabled from command line

Booting with pcie_aspm=off kernel parameter should disable ASPM altogether, so you might want to give it a shot...

lkraav · 2011-05-07 21:25:53

bisect successful, mine is caused by acer-wmi: https://bugzilla.kernel.org/show_bug.cgi?id=34682

Arch Linux

#1 2011-04-22 16:19:22

Suspend broken after upgrade to 2.6.38 on Lenovo L512

#2 2011-04-22 22:18:43

Re: Suspend broken after upgrade to 2.6.38 on Lenovo L512

#3 2011-04-29 18:59:06

Re: Suspend broken after upgrade to 2.6.38 on Lenovo L512

#4 2011-04-29 19:25:02

Re: Suspend broken after upgrade to 2.6.38 on Lenovo L512

#5 2011-05-02 08:39:48

Re: Suspend broken after upgrade to 2.6.38 on Lenovo L512

#6 2011-05-07 10:53:31

Re: Suspend broken after upgrade to 2.6.38 on Lenovo L512

#7 2011-05-07 14:12:29

Re: Suspend broken after upgrade to 2.6.38 on Lenovo L512

#8 2011-05-07 15:43:53

Re: Suspend broken after upgrade to 2.6.38 on Lenovo L512

#9 2011-05-07 21:25:53

Re: Suspend broken after upgrade to 2.6.38 on Lenovo L512

Board footer