ACPI issues

VisionIncision · 12-30-2011, 02:20 PM

Hi all,
I am currently having some issues putting my laptop into Suspend mode via XFCE.

Sometimes the laptop does not suspend. Sometimes it will suspend up to 5 minutes after closing the lid/clicking suspend.

Regards

business_kid · 01-01-2012, 01:26 PM

Snap! Same Problem by times.

Run acpid with -l (hack /etc/rc.d/rc.acpid)
You can then run
tail -fn 30 /var/log/messages

in a terminal and press acpi type things and see what's going on.

VisionIncision · 01-03-2012, 10:48 AM

Quote:

Originally Posted by business_kid

Snap! Same Problem by times.

Run acpid with -l (hack /etc/rc.d/rc.acpid)
You can then run
tail -fn 30 /var/log/messages

in a terminal and press acpi type things and see what's going on.

Thanks, I shall have a look at my logs when I am back on my own machine.

Regards,
Jack.

business_kid · 01-04-2012, 04:06 AM

I'm discovering other stuff and would like to know their function. /usr/lib64/pm-utils is full of stuff. There's a very informative README about this in /usr/doc/pm-utils-1.4.1/README.SLACKWARE and a maintainer who has got *zero* emails, so I have fixed that problem for him and posted him this thread link.

EDIT: Try the utilities pm-suspend and pm-hibernate. For me
pm-suspend seems to do the business, but fails to turn things off.
pm-hibernate actually hibernated! I was amazed.

rworkman · 01-04-2012, 08:46 AM

I just replied to the email, but to summarize it, suspend/hibernate problems are rarely (if ever, these days) a problem with the pm-utils package. The kernel is the most important piece of the puzzle there.

business_kid · 01-04-2012, 01:15 PM

The kernel is the vmlinuz-huge-2.6.37.6 with slackware 13.37, so you know more about it that I do. I will rebuild, but not this week.

I seem to be sorted in a half-assed way:
/usr/sbin/pm-hibernate hibernates in an acceptable timeframe
/usr/sbin/pm-suspend suspends but fails to turn off (ugh!)

but 'echo mem > /sys/power/state' does it ok. So I modified the hibernate script

echo disk > /sys/power/state gives delays of apparently random length after blanking the screen, cutting out the keyboard, but before writing anything to disk

The box has also developed an ugly habit of going into continuous restore-->hibernate cycles linked so some faulty stuff being written onto the drive. I boot slamd64 to get out of it, and it finds loads of errors recovering the journal. All very weird. I'll get out all the quirks you guys compile in to the kernel and that will probably sort things.

business_kid · 01-07-2012, 07:55 AM

Kernel 2.6.37.6 rebuilt largely from a previous attempt, and some issues remain.I have run down a few things, and come up with a question.

With acpid -l, I would get output in the logs of a hibernate command (my hibernate script), the first command of which was a 'sync;'. The next lines would say

Code:

Jan  6 18:54:14 harriet kernel: [21139.567385] PM: Syncing filesystems ... done.
Jan  6 18:54:14 harriet kernel: [21140.000319] sd 0:0:0:0: [sda] Synchronizing SCSI cache

I have figured that the "Sync" log entry is the Restore, and that it doesn't actually record the hibernate at all. Sorry for any wasted grey cells this may have caused.

The question involves this log segment from /var/log/syslog. Whassup?

Code:

bash-4.1$ grep blocked /var/log/syslog 
Dec 31 16:01:30 harriet kernel: [ 7920.624087] INFO: task kworker/u:5:1043 blocked for more than 120 seconds.
Dec 31 16:01:30 harriet kernel: [ 7920.624406] INFO: task pm-hibernate:3459 blocked for more than 120 seconds.
Dec 31 16:01:30 harriet kernel: [ 7920.624593] INFO: task kworker/u:3:3670 blocked for more than 120 seconds.
Jan  1 17:56:14 harriet kernel: [11880.621153] INFO: task echo:3553 blocked for more than 120 seconds.
Jan  1 17:56:14 harriet kernel: [11880.621373] INFO: task kworker/u:0:3554 blocked for more than 120 seconds.
Jan  1 17:56:14 harriet kernel: [11880.621614] INFO: task kworker/u:9:3561 blocked for more than 120 seconds.
Jan  1 18:10:28 harriet kernel: [12840.621131] INFO: task sync:3837 blocked for more than 120 seconds.

Before you attack my kernel, these are from the Slackware huge 2.6.37.6 kernel. Going off to google and check up on several things.