mi wrote:
In short, i would propose to first get acpi 'S3' working when calling it manually; and then see if the system still hangs.
Not possible - see my previous post :-)
| easier--I'm just flabbergasted at the amount of time it took me to even | realize that my power management settings weren't right...)
No, that was the amount of time to "guess" that they weren't right - and you're still barking up the wrong tree.
| > Usually, I'd say this is the BIOS, then. What does it's config say ? | Where do I look for that?No experience with acer. Maybe entering BIOS with F2 or DEL before boot, and looking for something appropriate (typical timespans like 30m, 15m ... ) and if the whole thing is set to OFF. Eventually you save the changed config before leaving to boot on. If you're unsure, note the 'original' values (there's always a safe reset-to-defaults, anyway). It's good to say hello to the BIOS at least once in a lifetime ;-)
No better ideas myself. I think this is a good place to look.
However, i don't like my own idea no more. If acpi runs from your Windows, the BIOS settings can't be that bad.
Well, that's not correct :-( See, a lot of BIOSes have code that checks you OS in the DSDT. Some people have made ACPI much more functional by modifying the DSDT to tell ACPI that you're running NT when you're running Linux - but that would be entirely dependent on the particular machine and what else the DSDT does...and definitely not for the faint of heart (and potentially hazardous to your laptop)
Instead, i think it could be that: (http://xtrinsic.com/geek/code/XF86Config-4.txt) Section "Monitor" Identifier "Generic Monitor" HorizSync 28-49 VertRefresh 43-72 Option "DPMS" <<< EndSection For a try, you can hash (#) it out.
Oh, yes. Definitely drop DPMS first. I haven't a clue what DPMS can do _besides_ blanking the screen, but it can definitely do that. And more than one laptop can't come back from that.
There are 'backported' patches for kernels at least >=2.4.18. The higher, the more are included by default, i think. As i understood, acpi is targeted to be _really_ implemented in 2.6 ! But 2.5.6 would be much better than 2.4.2, i guess.
2.5.59 or higher by now. When going with the 2.5 series you pretty much have to keep grabbing the latest until one works - I'm using 2.5.24 because I needed 2.5 but that was the only one I could get to compile. Haven't tried since 2.5.40 or so, though.
The acpid package description says, unpatched at least => 2.4.7 is required, if i get it right.
All that means is that the /proc/acpi structure changed about then, so the current (6 month old, or so) acpid won't work with the older /proc/acpi
I'm offline for ten days now. Good luck !
Too bad for emma :-)