You are not logged in.
After the big update, my suspend to ram stopped working. I was using the stock kernel26 and all was good. Laptop (Dell 600m) went to sleep and woke up perfectly fine as long as i wasn't using a console framebuffer. Now it goes to sleep but doesn't wake up at all. This is the problem I've had with every other distro; except Ubuntu. I was very happy when it worked on arch, as it is the most useful feature for laptops.
Anyway my question is this: what all changed in the new kernel? I suspect it is the kernel that is borking it. Under normal circumstances I would have just rolled back to the old kernel, but I wasn't sure if that was a great idea, as, IIRC, mkinitrd switched to mkinitcpio (or whatever it is) and I'm not sure if that would cause any problems down the road with other packages.
Any help, either to fix it or to confirm that it won't be detrimental to go back to the previous kernel, would be greatly appreciated.
Thanks in advance,
/jhs
Open Toes; Open Mind; Open Source.
Offline
I have the same problem. It goes into suspend to ram, but won't come out (and it isn't just vbetool broken because I can't ctrl+alt+del out of it).
Anyway, I reverted to the previous kernel (2.6.16.20-1) and none of the dependency checking thought I should downgrade the nvidia or ipw3945 drivers (which was necessary). So I did that manually as well. The machine was still functional, but now it wouldn't even go IN to standby. So I'm thinking there are some other things at play.
Maybe something is wrong with the new xorg too.
Offline
Arch on laptops is gonna be pretty scarce if things like this keep happening.
Please stop complaining. Bugs in upstream packages is simply something we cannot prevent. We don't have wide testing resources, we depend on users using the [testing] repository, in which 2.6.17 existed for quite a while before moving and no users reported problems.
It's most likely that some change in 2.6.17 has broken this. People, could you report exactly what symptoms you see on resume,
what turns on?
do your hard drives spin up?
does your screen turn on, but not display anything or is it off completely?
do any hibernated apps resume? try leaving a music player or something.
what chipset do you use?
what kernel are you using?
what graphics do you have?
Offline
It might not be strictly a kernel problem. When I downgraded to 2.6.16, it hung on going in to standby (the screen didn't turn off either).
I tried to downgrade to the previous versions of xorg and udev as well but still couldn't get out of standby. And on going into standby from text mode (actually framebuffer), it still won't come out.
In the hibernate.log file, it apparently just stops after going into standby:
hibernate: [98] Executing CheckRunlevel ...
hibernate: [99] Executing DoSysfsPowerStateSuspend ...
hibernate: Activating sysfs power state mem ...
Then nothing else:
So something hangs before the hibernate-scripts get to do anything coming out of suspend to ram.
Offline
just wanted to say that suspend-to-ram stopped working for me as well after upgrade to 2.6.17-beyond, (on a lifebook p7010). The system fails to return from sleep both while running X and in console. I would say everything seems to get back up except the video, but that's not quite true. I use acpi_sleep=s3_bios which always brings the video back. And with 2.6.17 the backlight gets turned on. Harddrive definitely spins up again too.
I'm gonna investigate it more when I have time.
Offline
I'm having the same problem with suspend to disk (running hibernate from hibernate-script) after I updated to 2.6.17. The problem (in my case) seems to be that swap is not activated at boot time. If I try to hibernate before I manually turn on swap, the logs show that it finds no swap and bails out. When I do turn on swap, it hibernates, but after the reboot there is still no swap, so it goes into the normal boot process. Maybe it has something to do with me switching to mkinitcpio today?
Offline
@Knappout: We seem to have the exact same experiences. I noticed that resume works from commandline, however. When X is running things mess up though. Can you check if it's the same for you. I was suspending and resuming without any fb device.
Offline
Hmmm... Suspend & resume seems to work for me with 2.6.17ck1. I'm gonna do some more experiments. I'm running custom built kernels, since I'm having this weird problem with stock kernels & my touchpad.
I'll get back as soon as I can. Unfortunately, I have to leave in a couple of hours and I won't have an internet connection for a couple of weeks...
Offline
Alright, I've tried a few times with kernel26ck now, and suspend-to-ram + resume seem to work fine.
I guess the problem is somewhere in the beyond patchset...
Offline
Hi guys.
I have suspend to RAM and Disk working with the beyond kernel. Kind of. Suspend to disk doesn't like to wake up sometimes, and there doesn't seem to be any consistent reason. I thought it was due to being asleep for a long time, but it isn't (and I can think of no way it could be, in hindsight - there's no power anywhere).
So, I thought I'd try the CK kernel as well, as there were also some issues with suspend and framebuffers. But, I can't get the suspend to disk to work with the CK kernel. Any ideas? I get this error:
[root@wijipad thomas]# hibernate -F disk
hibernate: No suitable suspend methods were found on your machine.
hibernate: You need to install a kernel with support for suspending to
hibernate: disk or RAM and reboot, then try again.
That is the command that works with the -beyond kernel. 'disk' is the config file I'm using.
I'm running on a Thinkpad T40. I'm not really sure which hardware details are appropriate, as I'm guessing this is a kernel issue.
Any info that's needed I'll provide asap.
Peace
Offline
ck-sources do not include the suspend2 patch. Therefore, you have to use traditional suspend to disk: this method was known to be unreliable and worst than suspend2, but it is now much better (as fast as suspend2, a bit more uglier).
You can continue to use hibernate, but in the disk file you resort to you have to choose the corresponding suspend method (try 'hibernate -h' to find the relevant parameters).
I am able to suspend to disk with ck-sources here (although I configured it on my own, I see through abs that the default config supports suspend to disk).
Mortuus in anima, curam gero cutis
Offline
Cheers for that, I'll give it a try as soon as I get time.
t
Offline
Hmmm no luck.
It seems that using the CK kernel, I can suspend using the old suspend method, but I can't resume. It just boots normally. And yes, kernel options are correct.
With the Beyond kernel, the suspend/resume works if you resume quickly (within a couple of mins). If the machine is left suspended overnight, the resume fails. It recognises the suspend image, but cannot load it. I get stuck with a blank screen or the 'thawing CPUs' message, needing a hard reboot.
Any ideas? I cannot fathom why the failure is based on how long the machine is suspended - anyone else?
Thinkpad T40, 512M, 1024M swap, 35Gb.
I tried disabling compression etc in the image, as my swap is bigger than my RAM. That didn't help. I'm not sure what else to do.
Ultimately I can live without it, as suspend to RAM works for short suspends, and a reboot isn't that much of a pain for a long suspend. But still, being able to suspend all my work safely would be useful, especially if the battery was running down.
Cheers.
t
Offline
I have just upgraded my Arch64 to the stock 2.6.17-ARCH kernel (x86-64), and I am having the same problem: The laptop goes to sleep on Suspen-to-RAM, but does not wake up. I have a line in my script immediately after
echo -n "mem" > /sys/power/state
returns, which is supposed to log a wake-up message. This line is never reached (according to /var/log/acpid).
Elad
Offline
Have you tried using the hibernate-scripts?
Incidentally I'm not sure the stock kernel supports suspend. I think you need to do
pacman -Syu kernel26beyond
Don't forget to update your modules.
Offline
Hmmm no luck.
It seems that using the CK kernel, I can suspend using the old suspend method, but I can't resume. It just boots normally. And yes, kernel options are correct.
With the Beyond kernel, the suspend/resume works if you resume quickly (within a couple of mins). If the machine is left suspended overnight, the resume fails. It recognises the suspend image, but cannot load it. I get stuck with a blank screen or the 'thawing CPUs' message, needing a hard reboot.
Any ideas? I cannot fathom why the failure is based on how long the machine is suspended - anyone else?
Thinkpad T40, 512M, 1024M swap, 35Gb.
I tried disabling compression etc in the image, as my swap is bigger than my RAM. That didn't help. I'm not sure what else to do.
Ultimately I can live without it, as suspend to RAM works for short suspends, and a reboot isn't that much of a pain for a long suspend. But still, being able to suspend all my work safely would be useful, especially if the battery was running down.
Cheers.
t
Did you enable the vbetool calls in the config files? Try different combinations of them (obviously, you need to install vbetool )
Mortuus in anima, curam gero cutis
Offline
Try also radeontool, if you hav a radeon card. Afterwards, try to unload different combination of modules to find which one is the guilty.
On the susped2 website, you can find many tips on how to make these tests. The tips can be easily applied also to suspend1 and suspend to ram.
Then you will need to unload the guilty modules in the hibernate config files.
Mortuus in anima, curam gero cutis
Offline
Have you tried using the hibernate-scripts?
Incidentally I'm not sure the stock kernel supports suspend. I think you need to do
pacman -Syu kernel26beyond
Don't forget to update your modules.
The default kernel (as the ck one) supports suspend1 and suspend-to-disk, as you can see in the default config through abs
Mortuus in anima, curam gero cutis
Offline
Have you tried using the hibernate-scripts?
Incidentally I'm not sure the stock kernel supports suspend. I think you need to do
pacman -Syu kernel26beyond
Don't forget to update your modules.
Things worked just fine with the stock 2.6.16-ARCH kernel (the use of the word "stock" may be misleading, but I don't think Arch patches the "real" kernel).
I tried the hibernate script, which caused a kernel panic.
As for the beyond kernel, I'm not sure it is available for Arch64 (but I haven't looked it up yet). I will try to install 32-bit Arch and see what happens.
Offline
Some of the problems seem to be related to swap. When I boot, I get a failure message for activating swap. This happens both with the regular kernel and with the beyond kernel (I've switched to 32-bit).
Offline
More information:
I have recreated my swap partition with mkswap. After reboot, everything is set up correctly. Once the laptop reboots from hibernation, though, I can see that swap fails to initialise, and dmesg complains about the swap signature.
I'm guessing the hibernation process damages the swap space in some way.
Offline
Yeah I've had that happen a few times too, not sure why.
I haven't enabled vbetools because I'm using a Thinkpad, and all accounts seem to suggest the option should be turned off for these machines.
I'll read through the suspend dodgy modules.
Would it be easier to just unload all modules and suffer a slower resume?
Offline
I realised I didn't have RadeonTool enabled in my suspend to disk configuration. I'll try that now, see how things are faring when I wake up.
Night
Offline
OK, well I'm now awake, and I could NOT resume. WTF?
So, I don't know what the hell is going on. I'll try unloading all modules next time, see if it comes back to life after that.
Peace.
Offline
If yo have a thinkpad with a radeon mobility chipset, than you need to look at kernel bug 3022, use radeonfb (you need to configure your kernel to do so) and use the relative patch I maintained for some time. Read the comments in kernel bugzilla.
The patch is going to be included in mm and is being considered for vanilla inclusion in 2.6.18.
If your thinkpad model is included in the whitelist in the patch, may be that the patch helps you immdiately. Otherwise, try to activate it with the kernel boot parameter. If this helps, put a comment in kernel bugzilla so that your subsystem vendor/device IDs can be listed.
In thinkwiki you find wide discussions of suspension problems affecting thinkpads with radeon video cards.
Mortuus in anima, curam gero cutis
Offline