[SOLVED] Odd behaviour from acpi

business_kid · 12-20-2011, 11:41 AM

My old box died, and I'm using my laptop with an external screen, mouse and keyboard. The laptop is a HP Compaq 6715S (2core Turion @2Ghz with 3G ram, AMD/ATI RS690/SB600, running slackware-13.37, using OSS drivers)

1. Fn-F3 & Fn-F4 are video options but they are weird. Fn-F3 kills ALL video both outlets, and hangs the box. Power cycle required. I think it's supposed to send the screen to the external monitor and back, but it's not happening.

2. The first time I touch Fn-F4 I get video out on the vga plug. Any subsequent presses do nothing and irritate the monitor, which spouts "No signal" but continues running :-/.

3. Hibernate & suspend misbehave. Hibernate does hibernate - eventually. I usually just walk off and leave it. Suspend doesn't usually. In both cases, screen goes black and keyboard is unresponsive, but it keeps going, until the Bermuda Triangle lines up with the Twilight Zone or something :-P. We are talking 1 - 10 minutes, at a guess, This problem has appeared slowly, which is weirdest of all.

Any ideas on fixes/Investigations?

business_kid · 12-21-2011, 11:27 AM

/Progress.

The suspend & hibernate delays seem to be process related. If I suspend/hibernate immediately after a resume, it happens normally.

Fn-F3 is labelled for video, but it is doing a suspend :-/. It generates the 'button/sleep' event :-o. I have obviously misread it, or something in the BIOS is clearly AWOL. But as suspend wasn't suspending, just blanking the screen, that might explain why both screens blanked and I wondered what was (not) going on.

Given that suspend unhooks all the interrupts, and hibernate probably does likewise, I groped in the logs to see a time difference such as this
[QUOTE]/QUOTE]

business_kid · 12-21-2011, 11:42 AM

/Progress.

Fn-F3 is labelled for video, but it is doing a suspend :-/. It generates the 'button/sleep' event :-o. I have obviously misread it, or something in the BIOS is clearly AWOL. But as suspend wasn't suspending, just blanking the screen, that might explain why both screens blanked and I wondered what was (not) going on.

The suspend & hibernate delays seem to be process related. If I suspend/hibernate immediately after a resume, it happens normally. Given that suspend unhooks all the interrupts, and hibernate does likewise, I switched on logging in acpid and groped in the logs to see a time difference such as this
[QUOTE]Dec 21 15:22:51 harriet acpid: received input layer event "button/screenlock SCRNLCK 00000080 00000000"
Dec 21 15:22:51 harriet acpid: rule from 1541[82:82] matched
Dec 21 15:22:51 harriet acpid: notifying client 1541[82:82]
Dec 21 15:22:51 harriet acpid: rule from /etc/acpi/events/screenlock matched
Dec 21 15:22:51 harriet acpid: executing action "/etc/acpi/Decpower.sh"
Dec 21 15:23:59 harriet kernel: PM: Syncing filesystems ... done./QUOTE]

The first line is the event. There is an unexplained delay of over a minute that time (in bold). Once I have the sync, all happens quickly. Decpower.sh is as follows:

Code:

#!/bin/sh
# /etc/acpi/Decpower.sh
# 
# Hopefully Manage to suspend/hibernate to disk

# Skip if we just in the middle of resuming.
 test -f /var/lock/acpisleep && exit 0

sync; echo  disk > /sys/power/state

Hardly a complicated or ambitious effort. Taking out the 'test' line doesn't improve things.

Any ideas Now?

business_kid · 12-26-2011, 10:56 AM

Marking this solved and reposting outstanding issue.