Most recent kernel where this bug did not occur: 2.6.9 (other kernels are not tested) Distribution: Mandrake 10.0 Hardware Environment: Soltek SL75FRN3-L motherboard with "nForce2 Ultra 400" chipset Software Environment: 2.6.12.2, 2.6.13 kernels with gcc-3.3.2, gcc-2.95.3 Problem Description: After poweroff called and system have been shutted down the "wake on RTC" fails. System won't wake up. Kernel recompiled with APM works OK, shuts down and wakes up the system successfully. Steps to reproduce: setting "wake on RTC" time in BIOS, booting Linux, /sbin/poweroff, wait for "wake on RTC" time (+ about 1-2 minutes ;).
Created attachment 5969 [details] 2.6.12.2 kernel and other logs
Created attachment 5970 [details] 2.6.13 kernel and other logs (PM and ACPI debug options turned on)
Does it work under Winxp? And please provide the acpidmp output. Thanks!
Exactly how are you instructing the RTC to wake the system -- via /proc/acpi/alarm, or other means? Note that acpidump is available in the latest pmtools here: http://ftp.kernel.org/pub/linux/kernel/people/lenb/acpi/utils
Created attachment 6038 [details] O.K., here it goes (acpidump.log) WinXP? Right now I have no ability to find it out. Maybe later... But it worked on Win2k. I'm doing nothing to RTC. Except one thing: I'm entering the BIOS setup, going to "Power Management" menu and setting "Wake on RTC" -> "Enabled", "Day of month" -> "0", "Time" -> "7:00:13". That's it. No any other manipulations with kernel/proc/Linux/etc. Strange one more thing comes up: in BIOS "wake up" time is set to "7:00:13". But 'cat /proc/acpi/alarm' says "20:02:13"! Isn't it strange? I don't know from where it gets it. Because changing "wake up" time in BIOS to another time DON'T changes information in /proc! 'cat /proc/acpi/alarm' still says "20:02:13"! Weird stuff.
Created attachment 6041 [details] debug patch Could you please apply the patch and open all wakeup devices (echo 'MMAC' > /proc/acpi/wakeup as an example). What's the result?
I've applied patch, opened all devices in /proc/acpi/wakeup... But there are no changes, the same old story.
Hmm, does this still happen in a recent kernel?
please re-open if this is still a problem in linux-2.6.22.stable or later