Sudden random shutdown after upgrade to 2.6.33.3-2

dabd · 2010-05-11 10:07:08

After doing a full system upgrade my laptop suddenly and randomly shuts down completely with no warning.

Is anyone experiencing this?

hokasch · 2010-05-11 10:53:34

- any messages in the log files?
- downgrade to verify which update introduced it
- watch the temps - overheating?

dabd · 2010-05-11 12:38:46

I just installed lm_sensors to monitor cpu temp.
It happened again and just after I restarted the computer the cpu temp were reading around 80 C, so the shutdown might be caused by overheating, I think.

If it is the cause how can I prevent overheating?

Last edited by dabd (2010-05-11 12:39:10)

hokasch · 2010-05-11 13:16:41

No idea...
there is someone else with the same problem:
http://bbs.archlinux.org/viewtopic.php?id=96622

but I would not just go with the workaround - if this is really is a regression/bug in kernel power management, that is quite serious, and you should definitely follow up on it (do some more testing and file a bug report - in the end, you want it to be fixed).

Last edited by hokasch (2010-05-11 13:17:27)

steve___ · 2010-05-11 14:13:00

dabd wrote:

If it is the cause how can I prevent overheating?

80C seems kinda low, but the thresholds can be different for each model of CPU.

what does this say:

cat /proc/acpi/thermal_zone/THRC/trip_points

Are you using cpufrequtils to scale down your CPU?

Is there anything else in your logs?

If you can reproduce the error or if it is happening frequently than do as hokasch suggests and downgrade the kernel to help narrow down the issue.

Last edited by steve___ (2010-05-11 14:15:28)

dabd · 2010-05-11 15:07:03

I have no subdirs under /proc/acpi/thermal_zone and I am not using cpufrequtils.
I couldn't find anything on the logs either.

steve___ wrote:

dabd wrote:
If it is the cause how can I prevent overheating?
80C seems kinda low, but the thresholds can be different for each model of CPU.
what does this say:
cat /proc/acpi/thermal_zone/THRC/trip_points
Are you using cpufrequtils to scale down your CPU?
Is there anything else in your logs?
If you can reproduce the error or if it is happening frequently than do as hokasch suggests and downgrade the kernel to help narrow down the issue.[

hokasch · 2010-05-11 15:43:11

you should keep an eye on cpu temps, e.g. with gtkrellm. what cpu do you have?
definitely make use of frequency scaling, it will help running cooler.

Does the workaround in the linked thread above work for you in terms of bringing down the temperature?
What kernel version was installed before the update? Can you downgrade to that version, and see if a) the random shutdowns stop b) temps change c) fan activity change d) thermal_zone is in /proc/acpi

dabd · 2010-05-12 14:16:23

hokasch wrote:

you should keep an eye on cpu temps, e.g. with gtkrellm. what cpu do you have?
definitely make use of frequency scaling, it will help running cooler.
Does the workaround in the linked thread above work for you in terms of bringing down the temperature?
What kernel version was installed before the update? Can you downgrade to that version, and see if a) the random shutdowns stop b) temps change c) fan activity change d) thermal_zone is in /proc/acpi

Intel(R) Core(TM)2 Duo CPU P8700 @ 2.53GHz
I downgraded from 2.6.33.3-2 to 2.6.33.3-1 and the problem still persists.
I installed gkrellm and when I rebooted the temp was 76ºC so I suspect that when it crashed it would be maybe around 80-81.

Looking at my /var/cache/pacman/pkg I can see that the following kernels:

-rw-r--r-- 1 root root   662559 Sep 25  2009 kernel-headers-2.6.30.5-1-x86_64.pkg.tar.gz
-rw-r--r-- 1 root root 36631051 Oct 14  2009 kernel26-2.6.31.4-1-x86_64.pkg.tar.gz
-rw-r--r-- 1 root root 23138068 Apr 19 23:33 kernel26-2.6.33.2-1-x86_64.pkg.tar.xz
-rw-r--r-- 1 root root 23115212 Apr 30 02:48 kernel26-2.6.33.3-1-x86_64.pkg.tar.xz
-rw-r--r-- 1 root root 23115724 May 10 00:41 kernel26-2.6.33.3-2-x86_64.pkg.tar.xz

The problem first appeared with 26-2.6.33.3-2 and after downgrading to 26-2.6.33.3-1.
Is it worth the trouble to downgrade further to kernel26-2.6.31.4-1, or is it going to cause a lot of dependencies to be removed or even make the system more unstable?

steve___ · 2010-05-12 14:28:52

The problem first appeared with 26-2.6.33.3-2 and after downgrading to 26-2.6.33.3-1.
Is it worth the trouble to downgrade further to kernel26-2.6.31.4-1, or is it going to cause a lot of dependencies to be removed or even make the system more unstable?

Try 'pacman -U' and see if there are dependencies.

$ pacman -U /path/to/file/kernel26-2.6.31.4-1-x86_64.pkg.tar.gz

dabd · 2010-05-12 15:10:56

steve___ wrote:

The problem first appeared with 26-2.6.33.3-2 and after downgrading to 26-2.6.33.3-1.
Is it worth the trouble to downgrade further to kernel26-2.6.31.4-1, or is it going to cause a lot of dependencies to be removed or even make the system more unstable?
Try 'pacman -U' and see if there are dependencies.
$ pacman -U /path/to/file/kernel26-2.6.31.4-1-x86_64.pkg.tar.gz

I downgraded the kernel to kernel26-2.6.31.4-1-x86_64.
I think I found the reason for the shutdown: the fan stops and the temperature rises (> 90º C) until the hardware shuts down automatically.
Now I have to figure why the fan stops and doesn't restart.
Any ideas?

Last edited by dabd (2010-05-12 15:11:14)

steve___ · 2010-05-12 15:31:48

So you're still seeing the problem with kernel26-2.6.31.4-1-x86_64?

Using top and sorting by CPU (shift-p), is there a process that is taking off and causing your machine to overheat? My guess is a while loop without a sleep when starting dwm from .xinitrc... but it's just a guess.

dabd · 2010-05-12 16:05:11

steve___ wrote:

So you're still seeing the problem with kernel26-2.6.31.4-1-x86_64?
Using top and sorting by CPU (shift-p), is there a process that is taking off and causing your machine to overheat? My guess is a while loop without a sleep when starting dwm from .xinitrc... but it's just a guess.

Yes it also happens with 26-2.6.31.4-1.
I can overheat the cpu by playing a video on VLC. By the way the VLC also is showing some lagging which didn't happen before.

For now I added pc=noacpi to the kernel boot line on grub and this removed the lagging from VLC(?) but the temperature is still rising and once the fan shuts down it doesn't restart

Last edited by dabd (2010-05-12 16:10:54)

steve___ · 2010-05-12 16:17:10

Using top and sorting by CPU (shift-p), is there a process that is taking off and causing your machine to overheat?

dabd · 2010-05-12 16:27:09

steve___ wrote:

Using top and sorting by CPU (shift-p), is there a process that is taking off and causing your machine to overheat?

No there is no process causing the overheating. The problem is that once the fan stops it doesn't restart and if I am using some processor intensive app, like VLC, it just overheats.

Why isn't the fan restarting?

P.S. pci=noacpi didn't work in turning off acpi so I'm using acpi=off instead.

Last edited by dabd (2010-05-12 16:30:09)

hokasch · 2010-05-12 16:29:02

Good question.
Before anything else, I would try with a live cd to rule out a problem related to hardware.

P.S. pci=noacpi didn't work in turning off acpi so I'm using acpi=off instead.

come on! does it make your fan start again or not?

Last edited by hokasch (2010-05-12 16:34:24)

dabd · 2010-05-12 16:39:25

hokasch wrote:

Good question.
Before anything else, I would try with a live cd to rule out a problem related to hardware.

The fan is working so it must not be a hardware problem. Could it be some other package that was installed during the upgrade that caused ACPI to stop working properly?

The fan module is not available:

 modprobe fan
FATAL: Error inserting fan (/lib/modules/2.6.31-ARCH/kernel/drivers/acpi/fan.ko): Operation not permitted

Shouldn't this module be available for ACPI fan control?

dabd · 2010-05-12 16:44:07

hokasch wrote:

Good question.
Before anything else, I would try with a live cd to rule out a problem related to hardware.
P.S. pci=noacpi didn't work in turning off acpi so I'm using acpi=off instead.
come on! does it make your fan start again or not?

with acpi=off fan control is relegated to the BIOS so it starts.

Ranguvar · 2010-05-12 17:15:15

So, let me get this straight. VLC is lagging, and the fan does not restart. With acpi=off, both of these problems are fixed?
I would try a few other kernel versions if you can, see if you can find one that works so you can give the kernel devs more info.

Then, file a bug in the ACPI subsystem.
There may be some tools you can use to try to force the fan off or on via ACPI. I'd search for those (may already be on the system), and see if they work. Etc. Check the ACPI info for the fan, too, see if it says it's on when it's off. Etc. etc.

dabd · 2010-05-12 17:22:22

Ranguvar wrote:

So, let me get this straight. VLC is lagging, and the fan does not restart. With acpi=off, both of these problems are fixed?

Yes.

Ranguvar wrote:

I would try a few other kernel versions if you can, see if you can find one that works so you can give the kernel devs more info.
Then, file a bug in the ACPI subsystem.
There may be some tools you can use to try to force the fan off or on via ACPI. I'd search for those (may already be on the system), and see if they work. Etc. Check the ACPI info for the fan, too, see if it says it's on when it's off. Etc. etc.

Ok, I'll try this.

dabd · 2010-05-15 13:55:49

I upgraded today to kernel 2.6.33.4-1 and the issue with the fan not restarting seems to be gone. I ran 'mprime' to stress test the CPU and it went well.

However, /proc/acpi/fan is empty and 'sensors' doesn't return any fan readings.

How can I retrieve fan readings?

I am running a Toshiba laptop with a Phoenix BIOS so toshiba_acpi won't work here.

UPDATE: suddenly the problem came back :-( I don't get it...

Last edited by dabd (2010-05-15 14:32:23)

MadCat_X · 2010-05-15 16:03:54

Hi dabd,

I wonder if you've seen this thread (http://bbs.archlinux.org/viewtopic.php?id=96913). It would appear that our problems have few things in common. Have you tried using omnibook module from SVN? I didn't work for me the latest kernels, but you might have better luck.

dabd · 2010-05-15 16:55:35

MadCat_X wrote:

Hi dabd,
I wonder if you've seen this thread (http://bbs.archlinux.org/viewtopic.php?id=96913). It would appear that our problems have few things in common. Have you tried using omnibook module from SVN? I didn't work for me the latest kernels, but you might have better luck.

I just tried the omnibook module and it doesn't work either.

MadCat_X · 2010-05-15 17:13:26

It needs some fiddling around to work properly, but if you had SIGKILL message in dmesg after doing "modprobe omnibook", then I guess you're out of luck just as I am. I'm gonna try to boot Ubuntu 10.04 from the Live CD to see if the fan works there.

MadCat_X · 2010-05-15 18:16:37

As I expected, ubuntu makes no difference. I don't have a free partition I could install it on, so I didn't try the "acpi.power_nocheck=1" kernel option I found on ubuntu formus. However, I tried it in arch and to also doesn't help here.

Just for the record, I found something pretty intriguing about this problem. A lot of people seem to report that the problem appeared after an update or something, but it wasn't there from a get go. Ubuntu 10.04 runs atop 2.6.32 kernel, so if there was a regression, it would have to be something introduced in 2.6.32 or possibly 2.6.31. I've had this issue for about 6 months and I'm pretty sure that I was dealing with it even in 2.6.31. Bottomline: what the world triggers this low fan speed problem in the first place? I cannot be the BIOS (at least in my case) as there is only one update available for my Qosmio X300 which I installed the day I brought my laptop from the shop. I swear there was no fan issue for at least 2 months.

dabd · 2010-05-15 18:24:45

MadCat_X wrote:

As I expected, ubuntu makes no difference. I don't have a free partition I could install it on, so I didn't try the "acpi.power_nocheck=1" kernel option I found on ubuntu formus. However, I tried it in arch and to also doesn't help here.
Just for the record, I found something pretty intriguing about this problem. A lot of people seem to report that the problem appeared after an update or something, but it wasn't there from a get go. Ubuntu 10.04 runs atop 2.6.32 kernel, so if there was a regression, it would have to be something introduced in 2.6.32 or possibly 2.6.31. I've had this issue for about 6 months and I'm pretty sure that I was dealing with it even in 2.6.31. Bottomline: what the world triggers this low fan speed problem in the first place? I cannot be the BIOS (at least in my case) as there is only one update available for my Qosmio X300 which I installed the day I brought my laptop from the shop. I swear there was no fan issue for at least 2 months.

I also tried downgrading to 2.6.31, which was the kernel I had been using from Oct 2009 to April 2009, and the problem persists, so I don't understand what changes have caused this issue.

Arch Linux

#1 2010-05-11 10:07:08

Sudden random shutdown after upgrade to 2.6.33.3-2

#2 2010-05-11 10:53:34

Re: Sudden random shutdown after upgrade to 2.6.33.3-2

#3 2010-05-11 12:38:46

Re: Sudden random shutdown after upgrade to 2.6.33.3-2

#4 2010-05-11 13:16:41

Re: Sudden random shutdown after upgrade to 2.6.33.3-2

#5 2010-05-11 14:13:00

Re: Sudden random shutdown after upgrade to 2.6.33.3-2

#6 2010-05-11 15:07:03

Re: Sudden random shutdown after upgrade to 2.6.33.3-2

#7 2010-05-11 15:43:11

Re: Sudden random shutdown after upgrade to 2.6.33.3-2

#8 2010-05-12 14:16:23

Re: Sudden random shutdown after upgrade to 2.6.33.3-2

#9 2010-05-12 14:28:52

Re: Sudden random shutdown after upgrade to 2.6.33.3-2

#10 2010-05-12 15:10:56

Re: Sudden random shutdown after upgrade to 2.6.33.3-2

#11 2010-05-12 15:31:48

Re: Sudden random shutdown after upgrade to 2.6.33.3-2

#12 2010-05-12 16:05:11

Re: Sudden random shutdown after upgrade to 2.6.33.3-2

#13 2010-05-12 16:17:10

Re: Sudden random shutdown after upgrade to 2.6.33.3-2

#14 2010-05-12 16:27:09

Re: Sudden random shutdown after upgrade to 2.6.33.3-2

#15 2010-05-12 16:29:02

Re: Sudden random shutdown after upgrade to 2.6.33.3-2

#16 2010-05-12 16:39:25

Re: Sudden random shutdown after upgrade to 2.6.33.3-2

#17 2010-05-12 16:44:07

Re: Sudden random shutdown after upgrade to 2.6.33.3-2

#18 2010-05-12 17:15:15

Re: Sudden random shutdown after upgrade to 2.6.33.3-2

#19 2010-05-12 17:22:22

Re: Sudden random shutdown after upgrade to 2.6.33.3-2

#20 2010-05-15 13:55:49

Re: Sudden random shutdown after upgrade to 2.6.33.3-2

#21 2010-05-15 16:03:54

Re: Sudden random shutdown after upgrade to 2.6.33.3-2

#22 2010-05-15 16:55:35

Re: Sudden random shutdown after upgrade to 2.6.33.3-2

#23 2010-05-15 17:13:26

Re: Sudden random shutdown after upgrade to 2.6.33.3-2

#24 2010-05-15 18:16:37

Re: Sudden random shutdown after upgrade to 2.6.33.3-2

#25 2010-05-15 18:24:45

Re: Sudden random shutdown after upgrade to 2.6.33.3-2

Board footer