So, i have this 2008 netbook, with 1GB RAM and a Intel Atom
Tried Haiku and looks like that works flawlessly, but when im gonna poweroff/reboot the machine, it goes normal, all windows closes, dont appear new messages, ect.
But the machine doesnt poweroff/reboot, looks like it hangs, in poweroff the only thing that stays is a blue screen, same color of the wallpaper, but not the wallpaper, for reboot, blacks out the screen, but the system doesnt reboot, stays that way
I asked for some people and they said that probably was some problem with ACPI, but they didnt know exactly what was causing
At least hard resetting when holding the power button works
Searched, and the only thread with a problem looking like similar to mine, was more saying that what “fixed” was disabling ACPI. And i dont want that
This machine looks old enough that it may just not shut down unless the ACPI shutdown command is invoked from CPU 0. We have code to do that, but it’s disabled because that was causing system hangs on other machines. I think that’s due to a different Haiku bug, which is that if you disable CPUs sometimes the OS will just hang. If that got fixed, we could re-enable this, which would probably fix shutdown for you.
(Alternatively, now that we have support for thread CPU affinity, we may be able to use that instead of disabling all CPUs besides CPU 0, and fix the problem in ACPI that way. This may be the better solution anyway.)
Don’t shoot me if this is a dumb idea. Could he go into process controller and disable all processors except cpu 0 and then try shutting down to see if that works around the bug? If it works, then he could turn the cpu’s back on after reboot.
So, now that i saw your comment, in the process controller there is CPU 1 and 2, CPU 1 is the CPU 0? or it should be shown as CPU 0 in the process controller
I don’t know why ACPI always seems to get suggested as the main problem. ACPI is rarely the problem. We havn’t found or fixed any bugs related to ACPI in a long time.
A more scientific way to do things would be to narrow down the problem, disabling components, drivers and apps until the problem is identified.
It is more often something hangs in the shutdown process.
Sometimes it is the media kit, an FS or a third party app.
The ancient ACPI problem that keeps getting iterated:
On very old hardware ACPI needed to shutdown on the boot cpu (which is not guaranteed to be the same as cpu 0).
We did that for a while but since we had (or still have) some scheduling or inter cpu communication problems it caused a lot more hangs on shutdown on “modern” hardware As almost all hardware that require that became ancient ( think Asus EEE pc) and can’t even boot haiku it is very unlikely that this is your problem.
Disabling CPU’s before shutdown would be a way to test for this behaviour, as long as you figure out which CPU booted. Also trying if reset works properly is interesting as we fall back to reset by the keyboard controller doing the same as “CTRL - ALT - DELETE” if we fail ACPI reset.So if reset works it is less likely to be an app issue.
So, via the boot loader, im disabling multiple components, drivers, apps, servers, ect. Giving yet the same problem, poweroff stays in a blank blue screen, and reboot justs stays in a black screen and dont reboot
The only way where actually rebooted normally and by it self, was disabling ACPI, to poweroff needs to hold power button when appears the window that can safely reboot, if i try to shutdown
But since im no dev to have total certainty , there are logs that shows what happens during shutdown/reboot? Because in the syslog files, i didnt find anything related to shutdown/reboot or anything that looked uncommon,
And forgot to say about the CPU, i tried to see with the debug output during boot to see what CPU was using to boot up, so, what CPU is it? (idk if matter, because being 2 threads, disabling one or the other didnt do anything)
So, since there were no updates to this post, i decided to install Linux distro in it
Any Linux distro is being able to poweroff and reboot with no problems, BUT, every time that the system boot up
I see some interesting messages related to ACPI and ACPI BIOS Error and invalid ELF headers
I know that what i send it now was related to Linux, but idk if this could mean something to the problem Haiku was facing in this netbook for reboot and poweroff
It seems that this BIOS is not fully compliant and linux is using a workaround.
With no big hope, I’d check Acer site to see if there has been a BIOS update to fix that.
You can also check in a linux boot log or syslog what is happening later during ACPI initialisation. Are the lines about buggy BIOS also appearing there? Is there mention of a quirk used? etc.
