Arch Linux

arkay

Member

Registered: 2008-05-23

Posts: 79

rtc wakeup [Solved - sortof]

Hi everyone,

Due to a very unfortunate mishap (eh... user error).  I had to re-install my root fs last week.

In the process I moved to kernel 2.6.26, previously 2.6.25.

Under 2.6.25 I had rtc wakeup with mythtv running 100%.

Since the upgrade however I can't get it to work at all.

The following:

echo 0 > /sys/class/rtc/rtc0/wakealarm
echo `date '+%s' -d '+ 5 minutes'` > /sys/class/rtc/rtc0/wakealarm
cat /sys/class/rtc/rtc0/wakealarm
cat /proc/driver/rtc

Produces the RTC timer set as:

1221115872
rtc_time        : 06:46:12
rtc_date        : 2008-09-11
alrm_time       : 06:51:12
alrm_date       : 2008-09-11
alarm_IRQ       : yes
alrm_pending    : no
24hr            : yes
periodic_IRQ    : no
update_IRQ      : no
HPET_emulated   : yes
DST_enable      : no
periodic_freq   : 1024
batt_status     : okay

I then shutdown -h now but the wake time will come and go and the machine will not wake.  I've tried with bios RTC enabled/disabled.  Nothing seems to make any difference.
The part that concerns me is "alrm_pending    : no", should it not say yes?  I can't remember if it used to or not.

I've also changed rc.shutdown to not write the HWCLOCK on shutdown as is often required.  Also makes no difference.

Can anyone offer any thoughts?  Does it still work for you on the current kernel?

My hardware is an asus p4b800 deluxe if that is of any consequence.  I honestly don't know where to go to debug this as what I need to see is buried in the bios when the power is off!
 
Thanks for any assistance.

Cheers,

Arkay.

Last edited by arkay (2008-09-11 16:42:34)

omgwtfbyobbq

Member

Registered: 2006-07-04

Posts: 226

Re: rtc wakeup [Solved - sortof]

The output of cat /proc/driver/rtc is the same for me wrt alarm_pending and it wakes up fine here w/ 2.6.26. Are you using UTC time for the hardware clock (rc.conf)? If not I'd change that, change rc.shutdown back to where it was, and disable (it's always like this IME but it's only been for three different systems) the alarm in the bios.

arkay

Member

Registered: 2008-05-23

Posts: 79

Re: rtc wakeup [Solved - sortof]

Hi,

Thanks for the suggestions.  I'll give it a go.

With my bios though it only worked with RTC wake enabled, when I had it working.

My bios and rc.conf are both UTC but that shouldn't matter for the test run above.

rtc_time        : 06:46:12
rtc_date        : 2008-09-11
alrm_time       : 06:51:12
alrm_date       : 2008-09-11

echo `date '+%s' -d '+ 5 minutes'` > /sys/class/rtc/rtc0/wakealarm should work whether the OS is set to localtime or UTC.

Failing that though I think I might clone my root partition to another partition and see if I can downgrade my kernel, at least it will verify the fault.

From what little I can find there were some big changes made to the 2.6.26 kernel with respect to rtc.

Cheers,

Arkay.

arkay

Member

Registered: 2008-05-23

Posts: 79

Re: rtc wakeup [Solved - sortof]

Just in case anyone runs into this I resolved the issue by downgrading to the 2.6.25 kernel.

There's obviously some sort of regression in the 2.6.26 release with my motherboard.

Cheers,

Arkay.

omgwtfbyobbq

Member

Registered: 2006-07-04

Posts: 226

Re: rtc wakeup [Solved - sortof]

FWIW I've found that if my bios is set to local time the RTC wakeup will be off by the difference between it and GMT. Damn thing was finicky as hell.

xxxspuddy

Member

Registered: 2007-05-15

Posts: 57

Re: rtc wakeup [Solved - sortof]

I've encountered a similar problem with the rtc wakeup works perfectly with 2.6.25 but upgrading to 2.6.26 (just the single kernel package) it does not work as expected.  If the wakeup time is the same day, it will not work but if it is set to some point the next day (I havent tested any further into the future) it will wake up at exactly midnight UTC.  I'm sure that's significant somehow...

---

A temporary file is just a pipe with an attitude and a will to live.

jacko

Member

Registered: 2007-11-23

Posts: 840

Re: rtc wakeup [Solved - sortof]

I'm sure that's significant somehow...

to the right bug tracker that might be invaluable information, but I highly doubt it will be useful to the bug hunters, being posted here. Nor will the likely hood of this bug getting fixed unless someones notifies the right individuals to fix it.

http://kerneltrap.org might be a good starting point.