Most recent kernel where this bug did not occur: 2.6.23.1 (this is a regression). Distribution: Bluewhite64 12 (64 bit version of Slackware 12). Hardware Environment: Laptop Acer Aspire 1511LMi, Athlon 64 3000, nForce3 chipset Software Environment: None / hardware bug Problem Description: When I boot the laptop, then issue a shutdown request (from KDE main menu), the laptop shuts down and power off correctly. However, once it is off, it restarted as soon as I close the lid. So I have to reopen the lid to get access to the power button and force a power off by pressing it 5s. Then I can close the lid and it stays off. So it seems the ACPI power off code issue a "restart on lid" (and maybe other kind events), even if the laptop was not suspended but really shut down. For information, suspend to RAM seem to work nicely (I use the lid event to go to hibernation mode, and the opening of the lid restarts correctly, but it should do it only after a suspend). Steps to reproduce: It may be hardware / BIOS dependant, but simply shuts down a laptop with "shutdown -Ph 0" then close the lid, it would restart.
Hi, Christian, Please attach the acpidump and the result of "cat /proc/acpi/wakeup".
Created attachment 13323 [details] acpidump for 2.6.23.1 That is for comparison with a working kernel.
Created attachment 13324 [details] wakeup for 2.6.23.1
Created attachment 13325 [details] Now the dumps for 2.6.24-rc1
Created attachment 13326 [details] wakeup for 2.6.24-rc1
Just for info: As requested in http://bugzilla.kernel.org/show_bug.cgi?id=9221, I added CONFIG_POWER_SUPPLY and battery in my config. I re-did the test, the lid closure still restarts the computer (not related to that misconfiguration therefore).
An interesting new fact that can help: If I unplug the power cord of the laptop before shutting it down, then it doesn't restart when I close the lid. It restarts only when it was shut down with the cord plugged in, not when operating on battery. Quite curious, isn't it?
Since this is a regression from 2.6.23 to 2.6.24-rc1. Christian, can you do a git bisect to found out which patch caused this regression please?
How do you do a git bitsect please? I guess this would be only on ACPI patches?
Never mind, the problem has disappeared with 2.6.24-rc2. Don't know why, but I cannot reproduce it anymore, so all is fine.
OK. Close this bug and mark it as CODE_FIX. Christian, Please reopen it if you can reproduce this problem.