APCI GPE 0x6F Interrupt Storm under OpenBSD

[IgorPetruk]

Hi Intel community,

I have recently bought and Inter NUC Kit with Core i7 11th gen (BNUC11TNHI70L00). After installing an OpenBSD operating system the system worked extremely slow with ACPI kernel thread taking 100% of one core, apparently slowing down other threads due to locks, etc.

It was established that the cause it uncontrolled firing of APCI GPE 0x6F. https://marc.info/?l=openbsd-bugs&m=166506935729650&w=4

Nobody seems to know what 0x6F even means. It was later reproduced on Jasper Lake N5105. Same interrupt. Masking the interrupt helps. This is (at least) reproducible with the latest 67-69 builds of the BIOS for the product.

I realize it is not a hugely popular operating system, but it probably reproduces a hardware bug, which ideally the hardware should not have. Other operating systems probably have some mitigations (not fixes) that prevent uncontrolled interrupt storms. OpenBSD does not have one, at least for ACPI GPE, so now users have to deal with it via very costly custom kernel builds with custom non standard patches.  OpenBSD developers are hesitant to chase fixes for interrupt storms, as they are usually resolved by BIOS updates from the hardware manufacturers. Usually those happen on cheap rare hardware, but now it surprisingly  happens in the NUC line.

Maybe it is not a bug, but if not, it would be nice to know what 0x6F is to at least narrow it down to some features that maybe can be disabled in BIOS. Even if I tried many settings there and nothing helped. 

I hope it is a right place to report this, please let me know if there is a better place.

I do not know where to report such things. 

[Alberto_R_Intel]

IgorPetruk, Thank you for posting in the Intel® Communities Support.

In reference to this scenario, it is important to mention that OpenBSD was not validated by Intel® as an officially supported Operating Suyetsm for

your Intel® NUC 11 Pro Kit NUC11TNHi70L:
https://www.intel.com/content/www/us/en/support/articles/000005628/intel-nuc.html

Based on that and in order for us to provide the most accurate assistance on this matter, we just wanted to confirm a few details about your system:

Is this a new Intel® NUC?

When did you purchase it?

Was OpenBSD working fine before on this specific Intel® NUC without the APCI GPE 0x6F issue?

If yes, when did the issue start?

Did you make any recent hardware/software changes that might cause this problem?

By any chance, did you test the Intel® NUC using Windows* as Operating System?

If yes, did you notice any kind of problem?

Which BIOS version is currently installed on the Intel® NUC?

Does the problem happen at home or in the work environment?

Regards,
Albert R.

Intel Customer Support Technician

[IgorPetruk]

Hi,

Thank you for a quick response.

> it is important to mention that OpenBSD was not validated by Intel® as an officially supported Operating Suyetsm for

your Intel® NUC 11 Pro Kit NUC11TNHi70L

Yes, I do understand. It might still be a good feedback that this happens to Intel developers and a chance to understand the issue for myself and maybe find a workaround.

> When did you purchase it?

29-Sep-22 on mouser.ie

> Was OpenBSD working fine before on this specific Intel® NUC without the APCI GPE 0x6F issue?

This is my first install on this device. Yes, in general the operating system works fine. Now it works fine for me after manually masking the interrupt in the kernel. There is no feature like that in the OS, it was masked directly in the kernel source code by me. The issue was reproduced by other users on OpenBSD mailing lists, but I don't know the details of their hardware, aside that is was Jasper Lake N5105.

> Did you make any recent hardware/software changes that might cause this problem?

I am not aware of any. Updating BIOS across different version from 67 to 69 and back made no difference. 

> By any chance, did you test the Intel® NUC using Windows* as Operating System? If yes, did you notice any kind of problem?

No, both Windows and Linux work fine at the moment. Not sure about Windows, but Linux automatically mitigates interrupt storms. That is not a fix in hardware or firmware, where the problem remains, just a mitigation that removes a side effect. 

> Which BIOS version is currently installed on the Intel® NUC?

https://www.intel.com/content/www/us/en/download/19698/bios-update-tntgl357.html,  has 69 version now, but it was causing random issues booting, so I reverted to 67th. Both were not fixing the GPE interrupt problem.

> Does the problem happen at home or in the work environment?

Home environment.

Thank you,

Igor.

[Alberto_R_Intel]

Hi IgorPetruk, You are very welcome, thank you very much for providing that information.

Yes, of course, for Intel® all the feedback provided by all of our clients is very important. Based on that, I will send your comments to the proper department for them to be aware of your remarks in order to keep improving the customer's experience while using Intel® products.

Any other inquiries, do not hesitate to contact us again.

Regards,

Albert R.

Intel Customer Support Technician