Bug 115127

Sounds very familiar. You should use APM then. The problem is that many BIOS's
are buggy. I won't dismiss this bug anyway, but please provide as much
information about your hardware as possible and provide us the last 1000 lines
of your syslog:

tail -1000 /var/log/messages > syslog.txt

no. The T40 should be perfectly ACPI capable.
What you could try is:

boot, press F3 for extra bootoptions in grub menu, enter "init=/bin/bash"

in the shell you get, enter the following:

mount /proc # ("/" is on the "." key on a german keyboard)
modprobe button
cat /proc/acpi/event

check if the different ACPI buttons (lid switch, sleep, power) send events. If
this works, it is a userspace bug (powersaved, acpid,...). If not, retry it with
the boot option "ec_burst=0"

I have got the same problem with my T23 and SUSE 10.0 Beta4.
<Fn>-<F4> works with Suse9.3 and also with Beta2+3.

<Fn>-<F4> stopped working after I run an update from beta3 to beta4.

I followed your instruction to boot with init=/bin/bash

Not all of the <Fn> key combinations show up as an event (e.g. <Fn>-<F12> 
which should be suspend to disk does not show anything) but others (like <Fn>-
<F4> do) 

cat /proc/acpi/event
button/sleep SLPB 00000080 00000001
button/lid LID 00000080 00000001
button/lid LID 00000080 00000002

For me, too, the closing the lid will cause a suspend to disk (if configured 
that way). but <fn>-<F4> has no effect.

Fn-F12 has no ACPI-function on any thinkpad i have seen, but fn-f4 not working
is strange. We need to look into this.
Can you check with Beta4 if you get an event in /var/log/acpid (running a
normally booted system, runlevel 3 or 5) if you press fn-f4? If this is the
case, we have a problem in the communication between acpid and powersaved or
powersaved is ignoring the sleep button for some strange reason. If not, acpid
has a problem.

Holger, Thomas, Danny: can you check if the sleep button works on _any_ machine?
My sony does not have an ACPI sleep button to check :-(

uname -a
Linux ltmnb2 2.6.13-3-default #1 Mon Aug 29 19:48:23 UTC 2005 i686 i686 i386
GNU/Linux

"nothing" means: no output in /var/log/acpid and no functionality detected

[Fn]+[F1] nothing

[Fn]+[F2] nothing

[Fn]+[F3] 
With APM it turns off backlight, sound and turns off harddisk. with ACPI nothing
happens (Bug also exists on suse9.3)
[Sun Sep  4 13:45:08 2005] received event "ibm/hotkey HKEY 00000080 00001003"
[Sun Sep  4 13:45:08 2005] notifying client 5715[0:0]
[Sun Sep  4 13:45:08 2005] notifying client 5744[0:0]
[Sun Sep  4 13:45:08 2005] notifying client 13871[0:0]
[Sun Sep  4 13:45:08 2005] completed event "ibm/hotkey HKEY 00000080 00001003"

[Fn]+[F4]
Does nothing, the configured sleep state will not be activated.
[Sun Sep  4 13:47:01 2005] received event "ibm/hotkey HKEY 00000080 00001004"
[Sun Sep  4 13:47:01 2005] notifying client 5715[0:0]
[Sun Sep  4 13:47:01 2005] notifying client 5744[0:0]
[Sun Sep  4 13:47:01 2005] notifying client 13871[0:0]
[Sun Sep  4 13:47:01 2005] completed event "ibm/hotkey HKEY 00000080 00001004"

[Fn]+[F5]
enables bluetooth, Popup appears "New hardware found" (I don't know if it works
correctly, I have no bluetooth device). Messages in /var/log messages, but no
messages in /var/log/acpid

[Fn]+[F6] nothing

[Fn]+[F7]
Should enable external VGA. I have no external display available at the moment
to see if it happens something. No output in /var/log/acpid

[Fn]+[F8] nothing

[Fn]+[F9] nothing

[Fn]+[F10] nothing

[Fn]+[F11] nothing

[Fn]+[F12] 
[Sun Sep  4 13:53:04 2005] received event "ibm/hotkey HKEY 00000080 0000100c"
[Sun Sep  4 13:53:04 2005] notifying client 5715[0:0]
[Sun Sep  4 13:53:04 2005] notifying client 5744[0:0]
[Sun Sep  4 13:53:04 2005] notifying client 13871[0:0]
[Sun Sep  4 13:53:04 2005] completed event "ibm/hotkey HKEY 00000080 0000100c"

ok. Got it. ibm-acpi is reconfiguring the sleep button. Bad.

the only thing you can do right now is either:
- create an hotkey script that triggers suspend (look in
/usr/share/doc/packages/powersave/contrib/acpi_hotkeys_* for examples)
- remove the ibm_acpi module, but this will cost you the ability to control the
display brightness via powersaved / software, hardware control should still work.

to prevent the ibm_acpi module from being loaded on next boot, you have to
remove it from /var/lib/acpi/laptop_modules.

no blocker, but not nice.

to #4: works for me with Acer Travelmate C300XMi as expected.

I can confirm that disabling ibm-acpi activates the functionality of [Fn][F4] on
a IBM T40.
[Fn][F3] still has no function :-(

What ibm-acpi is good for? Display brightness works as usual: [Fn][Home/End]
works and when AC is unplugged, the brightness is decreased. (and in kpowersave
I have not found something to adjust the brightness)

Fn-F3 is not supposed to do anything IIRC
ibm-acpi provides you with all those hotkeys - but with a different interface to
the sleep button :-( It also provides various functions for docking / undocking
and other Thinkpad specific stuff. If you don't know what it's good for, you
won't miss it :-)

Thank you for the information. I have no docking station, so I won't miss
something ;-)

Fn-F3 is intended to turn off the display backlight (it works with APM)
http://www-307.ibm.com/pc/support/site.wss/document.do?lndocid=MIGR-50023

On [F3] Key: a Display symbol and a "X" is printed on it
On [F4] Key: a Display symbol and a half-moon is printed on it
(see http://www.ixbt.com/portopc/ibm/t40/ea-keyboard2.jpg)

From comment #1 (suspend button does not work)
and comment #3 (suspend button does work with init=/bin/bash)
I expect that ibm_acpi module is disabling the sleep button?

If this is valid for all (or a lot of) Thinkpads we could treat this ACPI event:
[Sun Sep  4 13:47:01 2005] received event "ibm/hotkey HKEY 00000080 00001004"

simply as a button sleep event in the powersave daemon. This is an easy
workaround (in fact not even a workaround, but a fix).
Can someone confirm that this is always the sleep button on ibm_acpi machines?
Bad that this comes that late, I think the event should be the same on all
Thinkpads, but it seems to come from BIOS deeps and it cannot be said from the
ibm_acpi module...

up to beta3 it worked as expected. 
Isn't it better to find out what has changed between beta3 and beta4 and if this
change is really necessary?

we are loading ibm_acpi in beta4, we did not in beta3.

Thomas: this is no easy change in powersaved, since hotkey events are "other",
not "button", this logic would be quite bat.

But i found out, that ial handles the ibm_acpi events, so you can use ial to
assign "powersave -U" to them (and to all your other fancy IBM hotkey stuff).

It would be an easy change, we could simply filter out ibm/hotkey, check for the
1004 value and still let all others through as a an other event...

But you are right, this could conflict with the ial daemon...
Can this be enabled in the ial daemon by default?
Having another look at the module, I don't see why ibm_acpi overrides the sleep
button, seems to be an unexpected side effect.
So be aware that suspend could be invoked twice, but powersave daemon should be
able to handle this. If not we found a race condition bug :)

Timo/Seife: Will iald handle it?

iald is able to handle the hotkey events with prefix "ibm/hotkey". I think we should solve the issue by 
preventing ibm_acpi not to change the generic ACPI events. The events for buttons defined by the 
ACPI spec (Sleep Button, Power Button, Lid Switch) should _always_ be generated by button.c and 
thus have the generic "button/" prefix. The prefix "ibm/hotkey" is only fine for other ACPI events 
triggered on ThinkPads.

This is not easy...
It's probably not supposed by the ibm_acpi module to disable the sleep button.
Maybe the sleep button still works with ec_burst=0, can someone test that?
I don't see any relation between, the ibm_acpi module and the sleep button, so
this won't be fixable for 10.0.

Maybe those b.k.o bugs are related?

http://bugzilla.kernel.org/show_bug.cgi?id=5035
http://bugzilla.kernel.org/show_bug.cgi?id=4947#c5

The solution is almost done.

Will be fixed in RC2.

powersaved now recognizes the ACPI event "ibm/hotkey HKEY 00000080 00001004" as sleep button 
event.

Closing the bug. mbra, please verify.

I'll verify it, but when is RC2 released? (there is no information in the Roadmap)

it was only released internally - will be fixed in the final product.