Created attachment 50929 [details]
hwinfo.txt

Created attachment 50930 [details]
dmesg.txt

Looks to me like an ACPI-Problem. 

I heard of more machines with this problem.
Pavel do you know a solution for our 2.6.13 kernel and shutdown problems?

Eric Biedmann did some changes in that area. Does it also happen with 2.6.13
(vanilla)? Does it go away in 2.6.14-rcX?

I did not try. Where can I find the corresponding packages?

Shutdown works fine with vanilla-2.6.14-rc2

So it works with vanilla-2.6.14-rc2. What about 2.6.13?

Compal EFL30 has this bug with SL10.0 RC1.

Where can I find 2.6.13?

Could you try this one, please:

mount stravinsky:/myWork /mnt
cd /mnt/kernel_compiles/SL_10/linux-2.6.13_i386_reboot_fix
~trenn/bin/install_kernel i386

Edit /boot/grub/menu.lst according to the new installed kernel.

I placed an rpm in
ftp.suse.com/pub/people/trenn/10_0_kernel_reboot_fix/kernel-default-2.6.13-2.i586.rpm
(Could take a bit time until ftp mirrors are synced). Please let me know whether
it helps.

Maybe these bugreports are unrelated.
On Holgi's machine it seems to be the ibm_acpi module that causes the machine to
not switch off.
Loading the module and reading/writing to /proc/acpi/ibm is not enough.
It seems some interference between other ACPI module readings/writings.

We probably won't fix this for 10.0.
You have to delete the extra module in /var/lib/acpi/laptop_modules so that it
is not loaded.

Holgi could you double check, please and reopen if I just had pure luck.

Maybe it is something similar for the compaq, you could also try to avoid
loading the module by deleting any extra modules in /var/lib/acpi/laptop_modules.

We need the mainline solution as fast as possible for SLES10.
These extra modules are far too risky.
Timo, are you already working on something?
I vote for not including the brightness feature at all (this is the only feature
currently why we load all these broken extra modules?) if the generic module is
not usable until SLES10. Hmm, we should discuss that on mobile@suse.de.

I just tried to reproduce on a X32 which switched off just fine.

Thomas, "On Holgi's machine it seems to be the ibm_acpi module that causes the 
machine to not switch off." Is this an assumption or why are you writing 
"seems"? Have you tried whether the machine properly shuts down once the IBM 
module is not loaded?

Brightness is just one single option provided by the laptop modules. Please have 
a look on a ThinkPad's /proc/acpi/ibm/*.

Don't know what you are referring to with "all these broken extra modules".

I'll test the new kernel tonight if I can get to it.  Just for your info, my
laptop doesn't load any of the ibm modules since it isn't an IBM laptop.  I load
acpi modules appropriate for a Centrino based system.  It has a Pentium M 750,
Intel 915PM chipset, and ipw2200 wifi.  There is no fan or temp control
available so only the cpu-freq, battery, button are loaded.  I don't remember
other modules being loaded.  In SUSE 9.3 it was the battery module I needed to
not load since it didn't understand the battery.  The shutdown scripts would
shutoff the laptop with no problems but not reboot.  I could use suspend to ram
and suspend to disk.  With SUSE10 the battery module understands the battery and
I can use it.  The same modules are loaded plus the battery module, however, the
shutdown command can't shutoff the machine.  One quirk is the reboot command
will work in SUSE10 while it did not in SUSE9.3.

Thomas,
With the new kernel you provided, I can now shutdown with the power button using
the shutdown -h now command successfully.  I can not reboot with the shutdown -r
now command.  Suspend to disk works successfully.  Suspend to RAM works but
recovering takes several minutes.  I am not sure the battery capacity is being
interpreted correctly now either.  So this is a definite improvement!
Dow

acpi reports battery capacity correctly and the state under
/proc/acpi/battery/BAT1/state matches this.  Only the kpowersave applet is
showing an incorrect capacity of 1% when the battery is at 14%.  I could not
suspend to RAM using the powersaved config based on shutting the lid.  I used
"echo mem > /sys/power/state" and the laptop suspended fine.  So kpowersave and
powersave may be not functioning the way I'd like.  However, acpi seems to be
working much better.  I would like to be able to reboot though.  Let me know if
there is specific info I can provide at a higher debug level.  Thanks,
Dow

Dow: Please open another bug for that if you can reproduce these problems, this
one is long and mixed up enough.
This sounds like userspace, I expect Hal, possibly powersave daemon
misinterpreting ACPI values. You could increase the debug level of the powersave
daemon in /etc/sysconfig/powersave/common to 7 or even to 15 and have a look
whether you find anything suspicious.

Created attachment 53919 [details]
Reboot/Shutdown fix posted lately on acpi-devel by Eric W. Biederman (20.09.05)

Pavel you also followed up this one.
I think we should add it to 10.0, what do you think?
There seem to be a lot of HPs not powering off correctly, I hope most of them can shut down again with this one.
Please commit yourself or assign it to okir if you think it should go in.

Yes, the patch is probably good idea. It is in mainline these days...

*** Bug 117948 has been marked as a duplicate of this bug. ***

Thomas, what patches did the test kernels from comment #12 do? Are they
included in the 10.0 branch?

How is the patch from comment #22 related? The relevant change in that patch
seems to be that we now call the reboot notifiers when suspending to disk...
everything else is just rearranging code and adding comments. How does that
improve reboot/shutdown?

Thomas, you had kernel that fixed this issue (and then lots of other issues were added to this bugreport). Could you post a patch / commit it to suse kernel?

Olaf, you are right that #22 looks like cosmetics. It is in mainline, so it should be harmless.

Not only cosmetics, the little change(s) in the end seem to address and fix this shutdown issue (see comment #19, patch from #22 was the only patch I added to 10.0 branched kernel).
I cannot commit -> reassigning to Olaf.
Maybe the author also should be mentioned in the patch header:
by Eric W. Biederman (posted on acpi-devel - 20.09.05) (ebiederm@xmission.com)

Just to clarify: calling the reboot notifiers fixes the problem for you?
Or is it the addition of "system_state = SYSTEM_POWER_OFF;"? OR both?

I don't think we should take the whole patch - I've created a mininmal
patch that just adds the kernel_power_off_prepare() function and makes
power_down() call this instead of device_shutdown(). This patch will be
in the SL100_BRANCH in a few minutes.

Please verify the fixes the problem. Thanks!

Created attachment 55992 [details]
Minimal patch

This went into CVS

You can test by waiting some hours (better a day) until the new kernel is built, it is then placed here:
ftp.suse.com/pub/people/mantel/kotd/10.0-i386/SL100_BRANCH/kernel-default.i586.rpm
You can check with "rpm -qp --changelog ../kernel-default.i586.rpm |less" whether the changes are included into that kernel. Watch out for this string:

Mon Oct 31 09:12:37 CET 2005 - okir@suse.de
- patches.fixes/poweroff-fix: Call reboot notifiers of power off (118997).

Please reopen if it does not fix the problem anymore.

Neither not loading the ibm_acpi module nor using the new kernel 2.6.13-SL100_BRANCH_20051102110222-default fixes the problem. Machine does still not switch off.

It seems that you are facing a very specific Thinkpad A20 bug.


*** This bug has been marked as a duplicate of 113884 ***

Due to the fact that #113884 is fixed and I still have the same problem as well with current Kernel (2.6.13-15.7 default) -> reopen

Sorry Holgi, what you see seems to be a very machine specific bug.
Could you try the latest 10.1 installation. If you still have the problem there priority is high enough to look further at it, then please reopen. There were a lot ACPI changes recently and chances are high that it got fixed already and I don't have time to look at or backport already fixed stuff for a special machine.

Our HP ZE2113US laptop still not rebooting with SUSE 10.1 A3 but will turn off.

Does it reboot if you boot into runlevel 1 (just add a 1 as boot param).
If it still does not reboot, it is another bug than Holgi sees on his Thinkpad which seems to be somehow related to ACPI reads/writes (powersave daemon must be running).
If it still does not reboot, you could also try reboot=b (bios), reboot=c (cold), reboot=w (warm) and reboot=h (hard) and possibly reboot=s (smp) and tell us whether one/several of them works.
Could you open another bug for 10.1, please and assign it to pavel@suse.de (and take me into CC), I have too much bugs at the moment and I expect that it has another root cause than Holgis and the already fixed machine(s), there is already too much info mixed up in this bug report.

I have the same problem here , SUSE 10.0 , x86_64 , HP Compaq Presario
M2000.

I update Kernel with YAST(YOU) to 2.6.13-15.7-default but don't fix this annoying bug.

I read all the comments and i saw different problems here , to clarify my previous comment , my problem is that my laptop does not switch off after shutdown.

I also attach a hwinfo file.

Created attachment 61817 [details]
hwinfo.txt

Could you please open another bug report, there are too much different problems mixed up here.
I saw this one on a very similar machine (AMD Turion with similar chipset).
Please also try to shutdown in runlevel 3 (just add a 3 as boot param). If this does not work also try with 1. IIRC it only does not shut down if the X-server is running.
Please attach the hwinfo.txt and other useful information and assign the bug to me again and CC duwe@suse.de and pavel@suse.de.
Thanks.