ACPI causes kernel to hang at kernel_thread_helper+0x6/0x10

[abrondz]

Hi

I recently updated the UEFI on my brand new ASRock Z68 Pro3-M mainboard which is running Gentoo. After the update, I can no longer seem to run the latest gentoo-sources with ACPI enabled as it causes the system to hang at bootup with a kernel_thread_helper+0x6/0x10 error. The exact error message is quite long and I have no way of pasting it as this occurs before any logging services have been started. dmesg shows nothing either. However, the message is similar to one described here and here.

The error itself is quite vague, but after some research, I found that booting the system with acpi=off fixes the problem, unfortunately, it's a poor fix as disabling ACPI disables CPU functions such as hyperthreading. The only good solution so far for me has been to downgrade to vanilla-sources 2.6.38.8 which works like a charm, but I'd like to use gentoo-sources as that is the recommended kernel.

I've also tried pf-sources which seems to help with the problem at hand, but has its own issues.

I'm wondering what kind of issue this is, if it's a configuration issue on my behalf or a bug gentoo-sources or maybe even in the upstream kernel (the latest vanilla-sources show the exact same behaviour). What's also interesting is that all the descriptions of this problem are quite old, at least two years, meaning that this problem has been around longer than the 3.x kernel series. All in all, it leaves me very puzzled.

Anyone have any ideas on what this could be?

[khayyam]

[abrondz]

I would love to post the dmesg output, but as I mentioned, I'm not sure how to do it as the system crashes before the dmesg log gets written to disk. I'm not an expert user, so if I've missed something obvious, let me know

My .config can be found here: http://pastebin.com/FbD6gzCr

I'm using a 32 bit grub-legacy and a 32 bit system. I'm assuming it's MBR mode as haven't configured grub for anything else. Again, I'm not an expert in this, i simply followed the Gentoo handbook on this one.

[khayyam]

[abrondz]

Ah, I actually have that enabled and I do have a SysRq key, but looking at the Wikipedia link and trying out those key combinations gave little result as the system seems to be completely unresponsive.

[khayyam]

[abrondz]

Yeah, Alt + Print Screen *should* be SysRq, but it seems as if this is occurring way too early in the boot process, as you suggested.

[khayyam]

[abrondz]

Yeah, you're right, I didn't think that through..

I have to admit that I have been importing the kernel config from version to version. I didn't give it much thought at the time, but now that you mention it, maybe I should've started from scratch. Interestingly enough, I did configure the 3.x kernel first, then imported the .config into the 2.x kernel, meaning that whatever was breaking the 3.x kernel doesn't exist in the 2.6.x kernel.

I'm not using the nVidia framebuffer because it conflicts with the nvidia X11 driver, as mentioned here: http://www.gentoo.org/doc/en/nvidia-guide.xml#doc_chap3

I do understand that there is very little to go on here, in fact, the very reason I came here is because Google gave me nothing and the entire problem is leaving me very puzzled. That's also why I'm considering just sticking to 2.6.x, at least for the time being, as it seems to have everything I need. I will upgrade to 3.x as soon as possible, maybe this is something in the kernel itself, as mentioned earlier.

For the record, I did a few other tests of my own. I made a separate partition and did a quick Gentoo install from scratch, no old .config or anything, same problem. Both Arch Linux and Sabayon failed to boot, but interestingly, they crashed when attempting to launch a framebuffer. Ubuntu booted fine, but no virtual consoles were available, again suggesting a framebuffer issue. Finally, I tried using Ubuntu's .config in Gentoo, just for the heck of it, but it crashed with the same error as before (the one I have only been seeing in Gentoo).

What I can't make any sense of is why Gentoo crashes on ACPI initialization while all the other distributions crash much later, when the kernel was already fully loaded.

Sorry for the long, and probably very confusing, post. I'm quite confused myself..

[khayyam]

[abrondz]

Ah, I'll know that till next time then.

I tried switching over to the nvidia framebuffer, but no change, even disabling the framebuffer altogether does nothing. The only thing that has an effect on the situation is disabling ACPI in the kernel. I do understand that it doesn't mean that ACPI is to blame, but I'm long out of ideas on what else to do here.

What I mean, the system froze completely up, no response. The screen goes blank when launching the framebuffer and that's it. This is not what's happening in Gentoo, although the screen goes blank, it's only because the system has frozen up before anything gets displayed in the framebuffer. Removing the vga= in grub lets me see the the actual error.

I also tried getting rid of everything in the ACPI submenu, leaving only CONFIG_ACPI=y, but the problem was still there. I know that ACPI also enables certain other features, such as hyperthreading, maybe disabling these in turn could shed some light on the matter. Unfortunately, I don't know what all these features are.

[khayyam]

[abrondz]

Upgrading to gentoo-sources-3.4.4 actually fixed the problem, the system booted just fine. Unfortunately, nvidia-drivers did not compile under that kernel. When I tried gentoo-sources-3.3.8-r1, both the system and nvidia-drivers worked. All in all, a simple upgrade fixed the problem, it seems, and I now have ACPI under 3.x.

I don't quite know what happened in 3.2.x, but at least I have a working system now.

Thanks for all the help and suggestions

[khayyam]

abrondz ...

ahh good. As far as the nvidia-drivers are concerned, you could try those in ~arch or 302.17 which is package masked.

Again ... your welcome

best ... khay