ACPI problem -- not resuming from hibernate

Using MDK 10, 2.6 kernel, having some problems with ACPI on my Dell Inspiron 8500.

I attempted to use the hibernate function in KLaptop, it appeared to save the image to disk ok and shut off just fine. However, when powering up I have the problem.

It says it's resuming, prints some messages to that affect "PM: Restoring saved image."....no error messages at all, then gives the following message and promptly reboots:

PM: Waiting for DMAs to settle down.

After that it reboots. Does this everytime. Same if I boot into failsafe. So I cannot boot into linux at all, just a vicous cycle of rebooting......

Any ideas on how to solve this short of wiping out the system and starting from scratch would be appreciated.....

I guess I can boot up using the install CD as a rescue CD, right?? But even if I can, what exactly would I do to solve the problem? I don't have a clue......

Any ideas on how to solve this short of wiping out the system and starting from scratch would be appreciated.....

Everything else in ACPI worked fine (power management, battery, standby).......making me wish I hadn't tried to see if hibernate would work.....

Really all I need to know is for someone to tell me how to get it not to boot from the suspended image?

These are more of the messages it gives me, I can't get the whole thing because it flashes by so quickly:

/dev/hda6/
resuming from /dev/hda6
reading image
----------------------------------------------
hdc_start power step
hda_suspend
hda_complete resume request
PM: waiting for DMA to settle down


That's about it.

I was wondering if there is any boot file I can change to stop it from booting from there? Maybe set all the ACPI setting to =n in the /boot/config file??
Is there some way to purge the saved image it's trying to resume???

Please let me know

At the boot menu, add one of these options:
noresume
pmdisk=off
noresume2

If it is using grub, you can highlight the kernel and press e then you can edit the kernel and remove the resume line. (While in there look to see if a partition is specified. If so make sure it is your swap partition and that it is large enough to hold the image.)
Then hit b to boot (i think) Than it should boot normally. This is only temporary and to make the change perminat you have to edit /boot/grub/menu.list.
Hibanet works fine on my Compaq, so it does work provided the swap is large enough and the kernel is looking at the right partition.

Thanks for the help guys.

I'm using lilo as the bootloader, not grub.

I'm assuming you mean to edit the /etc/lilo.conf file with one of those options, for example:

image=/boot/vmlinuz
label="linux"
root=/dev/hda1
initrd=/boot/initrd.img
read-only
VGA = 768
append="noresume"
etc..................

or append="pmdisk=off" or =noresume2
correct?

It did use to say append="resume=/dev/hda6". hda6 is the swap partition, so that's correct, and it is 499MB......shouldn't that be big enough to hold the image?

Anyway, I've tried editing the lilo.conf file and it isn't working. I'm not sure if I'm editing the right lilo.conf file though. Why I think this is I've tried editing the label parameter, and the change to the name is never shown when I reboot. The /etc/lilo.conf file though has changed even after I reboot, so it's not reverting to the old file or anything like that.

So am I changing the right file? any other suggestions? Thanks!

Ok, I'm sorry that was a dumb question......I forgot you need to run lilo after you manually edit the lilo.conf file. I did that and it boot up fine without trying to resume.

Thanks a lot for your help!

When/If this happens again you can get around the reboot problem a bit quicker and simpler...

When Lilo boots giving you the boot options hit 'Esc' which will give you the lilo prompt and simply type...

linux noresume

... and hit enter. The system will then boot your default kernal passing it the noresume option. It will load the hibernation file but will then ignore it. Booting will continue as normal BUT be warned, ALL your file systems will not have been unmounted properly so it is worth doing the filesystem checks during the boot.

Also, the Swap partition will no longer work but there is a simple way around that as well... Just re-format it... This can be done from Mandrake Control Center >> Mount Points >> Partitions. Simply select the swap partition and click on format.

Once that is done as root type...

swapon /dev/hda6

will reactivate swap without having to reboot.

The failure to come out of hibernation and the continual reboots is a problem I have been fighting with for some time now with MDK9.2 and 10, I shall post a solution IF when I find one...

Made any progress on this issue? I'm sorta making a wish list of what Linux (Mandrake 10 specifically) would have to support for it to be a decent replacement for Windows XP. While Mandrake 10 is pretty cool, the list of things that it lacks *decent* support for, such as hibernation, is growing by the day. So much for desktop Linux, at least in my mind....

Anyway, if you have any new info, please share. Thanks :-)
- Joe

Hi all,

I am but a child when it comes to linux. I am hitting the same problem, basically it reboots when resuming from echo 4 > /proc/acpi/sleep, with the waiting... DMA... etc. This is on MDK10 download/community...

I *really* lust for hibernation on my laptop. It's an IBM thinkpad 390x, Pentium III, 256MB RAM.

Please, if anyone out there can help.. much appreciated.

Look into swsup. I don't think the 2.4 kernel can go into hibernation by itself.

Hi aaa,

Thanks for the help, I heard of swsusp but as a newbie I don't want to be compiling kernels and stuff, at least not at the moment.

AFAIK MDK10 is actually based on 2.6 something of the kernel so I *heard* it should all work out of the box.

If anyone is reading this and comes to the same problem we've been having, please I urge you, to drop by http://bugs.mandrakelinux.com/ and add your comment to the existing bug that someone has filed (and I have commented) regarding this issue - bug # 559

Peace out!

Mandrake 10 does indeed use the 2.6 kernel. So, in theory, hibernation should be about as good as it gets on Linux at the moment. However, this, as I understand it, is very much dependant upon the hardware you are using.

I did have hibernation working under 9.1 using swsusp and the 2.4 kernel, but have not been able to replicate this on 9.2 or 10 (Community or Official) using either swsusp OR the 'Klaptop hibernate option' (which do not seam to be cmpletely related).

I am not sure about the version 10 kernel (2.6.3) but the 9.x versions had the swsusp patch applied to the kernel and so, in theory, all you would have to do is sun the suspend.sh script (downloaded from swsusp.sourceforge.net and type hibernate at the command prompt (as root unless you setup sudo for the purpose). As stated above, this worked fine, with a little modification to the /etc/suspend.conf file, but I, as yet, have not managed to get it to work with 9.2 or 10 on the very same hardware.

One hint though... I found with 9.1 that it only worked if you disabled acpi (boot option) and did not run the acpid deamon at start up, thus running only APM.

I have, I think, tried all the combinations on 10 \now and have not had any luck. Further investigation required. and maybe I will need to compile my own kernel :-(

Hi all,

A compromise b/w my sanity and linux..

Thanks for all who are kind enough to post their findings.. just an update on my own situation.

* I somehow managed to figure out a way to get it to work (only tried it a couple of times) but to make a long story short I upgraded from community to official and 'forgot' the settiings I used!! #(@(! I didn't write it down b/c I thought it would be obvious.. It had something to do with nolapic and/or acpi=ht (or lack of) in lilo.conf. I tried different permutations but no success, BUT

* After browsing some more on the Net, I 'discovered' my thinkpad 390x does a suspend when you do FN F4 but you have to make sure apm (and its dependency suspend scripts) have to be installed! What can I say? I was looking for hibernate but having spent so much time on it, I've lowered my definition of success to getting suspend to work :> Only problem is that apparently all USB devices stop working when I turn it back on after a suspend, but luckily I don't really need USB in the field...

I've got a Dell Inspiron 8100, and I have to tell you that even when I had Windows on it, it wouldn't succesfully resume after hybernation. Nor with Linux. I therefore have concluded that its just the laptop.