IBM Thinkpad X41 suspend to disk unsatisfactory regression

[Christian Reis]

My X41's suspend to disk used to work perfectly in Breezy. I was convinced by one Matt Zimmerman to upgrade to the latest Dapper which is touted to be space-age technology. However, the suspend to disk functionality which I had grown to be so fond of no longer works in this version of Ubuntu.

How may I help further debug this concerning defect?

[Christian Reis (kiko)]

The problem I have, by the way, is that when I Fn-F12, the suspend process starts but never seems to finish, and requires shutting down forcefully (by holding down the power button for a long period of time).

Upon restarting Ubuntu nothing special seems to happen.

I have 512MB, and s-t-d didn't take very long on it in Breezy.

[Gaëtan Petit (gaetanp)]

i can confirm this on my Thinpad IBM X24

Tenshu

[Angelo Lisco (angystardust-gmail)]

I have an X40 and i noticed the same regression but only if my notebook is docked (th cd-rom is attached).

What's happend if you suspend to ram?

[Valentijn Sessink (valentijn)]

Please test first: press Fn-F12 and wait. If your laptop *does* hibernate (but it takes obscene amounts of time, like 5 minutes or more), your disk cache is the problem. There are a couple of solutions and I'm not sure which one is the best.
- /etc/hdparm could help
- /etc/laptop-mode/laptop-mode.conf can have "CONTROL_HD_WRITECACHE" on. Turn this off (set to 0). IMHO setting it off is stupid in all cases. It's meant to prevent data loss but the chance of losing data is IMHO much higher if your laptop just keeps running forever trying to suspend to disk.
(btw, for me, the Fn-F12 key doesn't work at all on my X40, though it did on Breezy, this is how I found your bug)

[Cristian Aravena Romero (caravena)]

What video card? Nvidia, Ati, etc.?

What's driver use (propietary or not propietary)?

I'm use video card ati, problem in hibernate and susspend with driver propietary.

Regards,
Caravena

[Cristian Aravena Romero (caravena)]

https://wiki.ubuntu.com/LaptopTestingTeam/CompaqPresarioV2417LA

[Paul Sladen (sladen)]

Kiko, is this still broken?

[Gaëtan Petit (gaetanp)]

yes it is.
I'm running my X24 under open drivers
i can hibernate with no problem (unlike under breezy)
but suspend still go to a infinite loop using my disk a full speed

...

[chris (chris-inferno)]

I have an X40 and suspend to ram hangs forever, but when deactivating the CD-Rom before suspend, it works.
I do this with "/usr/bin/hotswap -c 1 unregister-ide"

[Angelo Lisco (angystardust-gmail)]

i tried to compile 2.6.16 with the same .config of 2.6.15-18-386: both suspend-to-ram (sleep) and suspend-to-disk (hibernate) work again when my x40 is connected to an X4 ultrabase docking station (with cdrom)...
i know 2.6.16 will not be included in dapper so maybe we can ask Ben Collins to backport some bits from 2.6.16 (IMHO it related to the ide subsystem).

[Paul Sladen (sladen)]

This may also help with bug #6367.

[Eric Johnson (eric-el-studio)]

Valentijn's suggestion for turning off laptop-mode.conf's "CONTROL_HD_WRITECACHE"
seems to work on my Thinkpad X22 (radeon xorg driver).

It also significantly speeds restore from hibernate.

Any way to get this to be the default for laptop-mode in these machines?

[Christian Reis (kiko)]

Fn-F12 no longer does anything for me on this laptop (as of a dist-upgrade to T-5 minutes). Fn-F4 does suspend to ram successfully. Should I try and trigger this manually?

[Paul Sladen (sladen)]

Can you:

  sudo tail -0f /var/log/acpid

while pressing Fn-F12 and also paste the output of:

  cat /proc/acpi/ibm/hotkey

then if you're seeing the event, check that it is making it as far as HAL with:

  lshal -m

and if it's getting that far, this is maybe a gnome-power-manager issue.

[Christian Reis (kiko)]

On Mon, Apr 10, 2006 at 11:09:34PM -0000, Paul Sladen wrote:
> sudo tail -0f /var/log/acpid

[Wed Apr 12 14:32:33 2006] received event "ibm/hotkey HKEY 00000080 0000100c"
[Wed Apr 12 14:32:33 2006] notifying client 3914[120:120]
[Wed Apr 12 14:32:33 2006] executing action "/etc/acpi/hibernatebtn.sh"
[Wed Apr 12 14:32:33 2006] BEGIN HANDLER MESSAGES
[Wed Apr 12 14:32:33 2006] END HANDLER MESSAGES
[Wed Apr 12 14:32:33 2006] action exited with status 0
[Wed Apr 12 14:32:33 2006] completed event "ibm/hotkey HKEY 00000080 0000100c"

> cat /proc/acpi/ibm/hotkey

status: enabled
mask: 0xff9f
commands: enable, disable, reset, <mask>

> lshal -m

platform_i8042_i8042_Kbd_Port_logicaldev_input condition ButtonPressed = hibernate

Now what?

[Matt Zimmerman (mdz)]

Please don't hijack this bug to discuss the problem with the hibernate key (bug #39240); you can use "sudo /etc/acpi/hibernate.sh" to test meanwhile

[Zack Cerza (zcerza-deactivatedaccount)]

I just now saw the same behavior noted in the first comment to this bug. 2.6.15-19-686.

[Angelo Lisco (angystardust-gmail)]

I had the same problem before uploading to 2.6.15-21...
the last kernel upload (2.6.15-21) fixes many things regarding the ata/ide support...

(from the changelog)
  * libata: Backport of ACPI handling patch.
   * ide-acpi: Replaces our old IDE ACPI support code with new and shiney
     stuff that should work better on some hardware.
   * acpi-osl: Fixes some cases where code could sleep during resume (not
     allowed, since interrupts are disabled).
   * acpi: Bad characters shouldn't be fatal.

Ben Collins and Matthew Garrett ROCK! Now i can suspend-to-ram and suspend-to-disk again when my x40 is docked!!!

please, try again after an 'apt-get dist-upgrade'...so maybe we can finally close this old bug!

[Matt Zimmerman (mdz)]

The original problem reported in this bug is no longer present, and a secondary problem (missing swap partition) was corrected during an IRC debugging session. All is well.