[ltp] ACPI sleep works firts time

[linux-thinkpad@linux-thinkpad.org]

On Fri, 20 Aug 2004, Christopher Hubbell wrote:

> I've finally made some progress with getting my T23 to sleep.  I've 
> created a /etc/acpi/events/lid.conf which calls a simple shutdown script:
> 
> rmmod uhci_hcd
> rmmod ehci_hcd
> echo "mem" > /sys/power/state
> modprobe ehci_hcd
> modprobe uhci_hcd
> 
> If I run the script manually, all is well.   I close the lid and the 
> laptop shuts down as I expect.  I can then use the power button to bring 
> it back from its initial sleep, but from this point on the acpid doesn't 
> seem to trap events as lid closings no longer send the laptop to sleep.  
> I can still manually issue the command, but acpid doesn't seem to be 
> able to after its first run.  I'm using FC2 / 2.6.8-1.521.  Any 
> sugestions would be greatly appreciated.

If I can encourage you, I had the same symptoms on my T40 with an
earlier FC2 kernel, and got round them by compiling my own kernel.
I did this with some reluctance, but found it was the only way.
I'd recommend the latest from kernel.org (2.6.8.1) plus the latest
ACPI patches for it from http://acpi.sourceforge.net/ (top right on
the page).

Also, you'll need the kernel.org kernel if you want SWSUSP
(suspend-to-disk).  The FC2 kernel source RPM (kernel-sourcecode)
doesn't compile correctly if you enable CONFIG_SOFTWARE_SUSPEND in
it, and it's disabled in the binary kernel.  SWSUSP is really worth
it over BIOS suspend-to-disk - infinitely quicker once working.

I'd also strongly advise using ibm-acpi with this setup.

Finally I'd mention that most accounts still use the older /proc
method to sleep: so the equivalent to your echo above would be:

echo 3 > /proc/acpi/sleep

(4 for suspend-to-disk). Haven't a clue if this makes a difference.

Hope this helps!