System doesn't power off with ACPI

Bug #9185 reported by bluegeek
12
Affects Status Importance Assigned to Milestone
linux-source-2.6.15 (Ubuntu)
Fix Released
Medium
Herbert Xu

Bug Description

When I shutdown my computer (Shuttle SK41G) the system doesn't Power Off.

Last message is: acpi_power_off called

Trying to boot with nolapic ...
(grub) kernel /boot/vmlinuz-2.6.8.1-3-k7 root=/dev/hda6 ro quiet splash
noapic
... the systems doesn't boot:
BIOS bug, local APIC #o not detected!
Starting Ubuntu...
(but it doesn't start)

Booting with noapic doesn't solve the problem.

Revision history for this message
bluegeek (bluegeek) wrote :

Created an attachment (id=509)
dmesg output

Revision history for this message
Phil Housley (undeconstructed) wrote :

Seconded; this happens on my shuttle too, and also on my normal PC. Both are
VIA chipsets, but very different ones.

Revision history for this message
Herbert Xu (herbert-gondor) wrote :

(In reply to comment #1)
> Created an attachment (id=509) [edit]
> dmesg output

Firstly you're using an old kernel with MSI turned on. Please retest with the
latest kernel. Secondly please use dmesg -s 32768 as your kernel messages were
incomplete. Thanks.

Revision history for this message
bluegeek (bluegeek) wrote :

(1)
There are no differences between dmesg -s 32768 and dmesg (at least in my system):
blueboy@ubuntu-pandora:~ $ dmesg -s 32768 > dmesg2.out
blueboy@ubuntu-pandora:~ $ dmesg > dmesg.out
blueboy@ubuntu-pandora:~ $ diff -s dmesg.out dmesg2.out
Files dmesg.out and dmesg2.out are identical

(2)
According to Synaptic I have installed the last version of the k7 kernel:
linux-image-2.6.8.1-3-k7 (2.6.8.1-16)

Let me know if you need more info. Thanks four your help.

Revision history for this message
Herbert Xu (herbert-gondor) wrote :

(In reply to comment #4)
> (1)
> There are no differences between dmesg -s 32768 and dmesg (at least in my system):

Never mind, the Ubuntu kernel's buffer is only 16K.

> (2)
> According to Synaptic I have installed the last version of the k7 kernel:
> linux-image-2.6.8.1-3-k7 (2.6.8.1-16)
>
> Let me know if you need more info. Thanks four your help.

Well your dmesg output contradicts this. Please do uname -a and tell me what it
says.

Revision history for this message
bluegeek (bluegeek) wrote :

Created an attachment (id=545)
dmesg output

Revision history for this message
bluegeek (bluegeek) wrote :

Attached dmesg output again.

blueboy@ubuntu-pandora:~ $ uname -a
Linux ubuntu-pandora 2.6.8.1-3-k7 #1 Tue Oct 12 14:12:34 BST 2004 i686 GNU/Linux
blueboy@ubuntu-pandora:~ $ ls /boot/ -l
total 6476
-rw-r--r-- 1 root root 56107 2004-10-12 15:09 config-2.6.8.1-3-k7
drwxr-xr-x 2 root root 4096 2004-10-19 23:35 grub
-rw-r--r-- 1 root root 4476928 2004-10-14 22:21 initrd.img-2.6.8.1-3-k7
-rw-r--r-- 1 root root 81732 2004-10-11 12:06 memtest86+.bin
-rw-r--r-- 1 root root 843624 2004-10-12 16:21 System.map-2.6.8.1-3-k7
-rw-r--r-- 1 root root 1134936 2004-10-12 16:21 vmlinuz-2.6.8.1-3-k7
blueboy@ubuntu-pandora:~ $ dpkg -l linux-image*
Desired=Unknown/Install/Remove/Purge/Hold
| Status=Not/Installed/Config-files/Unpacked/Failed-config/Half-installed
|/ Err?=(none)/Hold/Reinst-required/X=both-problems (Status,Err: uppercase=bad)
||/ Name Version Description
+++-=========================-=========================-==================================================================
un linux-image <none> (no description available)
un linux-image-2.6 <none> (no description available)
pn linux-image-2.6-386 <none> (no description available)
ii linux-image-2.6-k7 2.6.8.1-13 Linux kernel image for
version 2.6 on AMD K7.
pn linux-image-2.6.8.1-3-386 <none> (no description available)
ii linux-image-2.6.8.1-3-k7 2.6.8.1-16 Linux kernel image for
version 2.6.8.1 on AMD K7.
pn linux-image-386 <none> (no description available)

Revision history for this message
Herbert Xu (herbert-gondor) wrote :

Thanks for the confirmation.

Please boot into a minimal environment by using the boot arguments
init=/bin/bash, Once there simply issue poweroff -f. Please let me know
whether this powers the machine off or not.

Revision history for this message
bluegeek (bluegeek) wrote :

No, it doesn't power off this way too :(

Revision history for this message
Herbert Xu (herbert-gondor) wrote :

Please boot the kernel at

http://gondor.apana.org.au/~herbert/ubuntu/acpi/vmlinuz-2.6.8.1-3-k7

and give me the last few messages after you power it off.

Revision history for this message
bluegeek (bluegeek) wrote :

last messages are:

acpi_power_off_called
LAPIC called
set CPU done

and it doesn't power off. I have windows XP installed in the same machine I can
try to find information there if you tell me where to find it.

Revision history for this message
Herbert Xu (herbert-gondor) wrote :

Thanks. Please try the updated image at

http://gondor.apana.org.au/~herbert/ubuntu/acpi/vmlinuz-2.6.8.1-3-k7.1

It has ACPI debugging enabled. Just before you power off, please echo
0xffffffff into /proc/acpi/debug_level and /proc/acpi/debug_layer. Please
double check that they're set correctly by catting them.

After that proceed to power off and give me the last few messages.

Revision history for this message
bluegeek (bluegeek) wrote :

Created an attachment (id=592)
shutdown screenshot

Revision history for this message
bluegeek (bluegeek) wrote :

Attached image with the state of the screen after the power-off. Thanks.

Revision history for this message
Gord Berta (gberta) wrote :

I had the same problem on my tyan 1854 mobo with a celeron 400 and 320 mb ram.

I edited the /etc/boot/menu.1st file as follows:

title Ubuntu, kernel 2.6.8.1-2-386 (recovery mode)
root (hd0,0)
kernel /boot/vmlinuz-2.6.8.1-2-386 root=/dev/hda1 ro single acpi=force
initrd /boot/initrd.img-2.6.8.1-2-386
savedefault
boot

By adding acpi=force it now turns off when I shut down.(In reply to comment #0)

Revision history for this message
Matt Zimmerman (mdz) wrote :

(In reply to comment #15)
> By adding acpi=force it now turns off when I shut down.(In reply to comment #0)

It sounds like you were having a different problem than the reporter, then. In
the attached dmesg, ACPI was being enabled, while in your case it was apparently
disabled by default due to the blacklist.

Revision history for this message
gela (gerald-orixa) wrote :

Created an attachment (id=680)
kernel reconfiguration made acpi poweroff work...

Same problem, vanished after reconfiguring and recompiling...
I don't know which of the numerous changes actually did the job, it may be
helpful anyway...

Revision history for this message
Gustavo Broos (gbroos) wrote :

I had the same problem on a VIA chipset, Abit VA-10.
It seems that the problem is due to the kernel patch which is applied for fixing
the same problem on IBM T42 machines.
I cannot find anymore the Ubuntu bugzilla bug on which the fix for the T42 is
accepted.
But this is the patch, the two lines that it adds break poweroff on my VA-10:

--- linux-2.6.7.bak/drivers/acpi/sleep/poweroff.c 2004-06-16 07:20:04.000000000
+0200
+++ linux-2.6.7/drivers/acpi/sleep/poweroff.c 2004-06-17 19:44:49.000000000 +0200
@@ -9,11 +9,13 @@
 #include <linux/init.h>
 #include <acpi/acpi_bus.h>
 #include <linux/sched.h>
+#include <asm/apic.h>

 static void
 acpi_power_off (void)
 {
  printk("%s called\n",__FUNCTION__);
+ disable_local_APIC ();
  /* Some SMP machines only can poweroff in boot CPU */
  set_cpus_allowed(current, cpumask_of_cpu(0));
  acpi_enter_sleep_state_prep(ACPI_STATE_S5);

Revision history for this message
Phil Housley (undeconstructed) wrote :

This is now working for me in Hoary's 2.6.9-k7 kernel.

Revision history for this message
Matt Zimmerman (mdz) wrote :

Sounds like this was fixed by:

linux-source-2.6.9 (2.6.9-2) hoary; urgency=low

  Changes by Fabio M. Di Nitto:
[...]
  * Update disable-lapic-in-acpi-power-off.dpatch with a portable fix from
    http://bugzilla.kernel.org/show_bug.cgi?id=3643.
    Add patch disable-lapic-in-acpi-power-off-2.dpatch.

Please reopen if someone is still experiencing this issue on the same hardware.

Revision history for this message
bluegeek (bluegeek) wrote :

(In reply to comment #20)
> Please reopen if someone is still experiencing this issue on the same hardware.

I solved this problem installing 2.6.9-1-k7 from hoary (but i'm still using warty)

Thanks to the Ubuntu people for your work.

Revision history for this message
Richard T. Bumby (bumby) wrote :

I have an Intel 845EBG2 board. The power button and other shutdown methods using ACPI have been working with Fedora Core 3, but I have not been able to complete the installation of Ubuntu 6.06. Everything has been copied to the disk, but the install process stopped at about the 80% point when intiallizing "power" (I don't have the exact message handy). Also, when I tried to shut down from the live desktop prior to attempting the install, the screen went blank but didn't power off.

Since I will need to continue using Fedora for current work until I get up to speed with Ubuntu, I hope to start Ubuntu from my existing grub bootloader. Configuration advice and request for test will be welcome.

Revision history for this message
Clive Darra (osde8info) wrote :

REAPPEARED on IBM Thinkpad T20 hangs on poweroff or shutdown from 6.10 live cd or 6.10 hd install

$ uname -a
Linux ubu 2.6.17-11-generic #2 SMP Thu Feb 1 19:52:28 UTC 2007 i686 GNU/Linux

$ lshal
  smbios.bios.release_date = '12/21/1999' (string)
  smbios.bios.version = 'IYET61WW (1.22 )' (string)
  smbios.bios.vendor = 'IBM' (string)

TEMP FIX
add acpi=force to grub kernel command line

see also bug 11821

To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.