NUC10i3 IRQ problem

[MKonr1]

Hello, I'm running NUC10i3 with Ubuntu 20.04 on board. I have a problem with cpu interrups causing issues with deeper CPU sleep and increased power usage. Also load is always 1 even if machine has nothing to do.

I made a reasearch and found that device named TPS6598x interrupts my CPU. This device is related with USB and according to datasheet it's "USB Interface IC USB Type-CG and USB PD controller power switch and high-speed multiplexer ". I have nothing connected to NUC except power plug and ethernet cable.

Screenshots: https://imgur.com/a/uw9NDCi

How to solve this issue?

Kernel version:

Linux NUC 5.6.11-050611-generic #202005061022 SMP Wed May 6 10:27:04 UTC 2020 x86_64 x86_64 x86_64 GNU/Linux

Bios version:

FNCML357 Version: 0039 Date: 3/12/2020

[User1589201909612646]

Hi there, facing probably the same issue on a NUC10i7FNK, even the irq is the same : irq/65-i2c-INT3      

I noticed one core is in constant use by checking top, I have a constant average load of 1.00 at least.

Have checked this with ArchLinux installed, a homebrew Live ISO and LinuxMint live ISO, all leading to this constant i2c irq beating the box.

Maybe the firmware or some buggy driver...

If someone at Intel could confirm they are working this issue?

Where can we drop a bug report or would this be enough?

Thanks

[Alberto_R_Intel]

MKonr1, Thank you very much for posting in the Intel® communities support.

How many units are affected?

What would be the impact of the Intel® NUC not working properly?

In reference to this scenario, in order to try to fix this problem, what we recommend first of all is to update the Kernel.

If the problem persists after that, please provide the SSU report:

https://downloadcenter.intel.com/download/26735/

And also de information showing when using the following command: /var/log/dmesg

To attach the report please select the "clip' icon on the lower-left corner.

Any questions, please let me know.

Regards,

Albert R.

Intel Customer Support Technician

A Contingent Worker at Intel

[n_scott_pearson]

I also notice that no one has brought up the possibility that I think has the highest probability of being the case, namely, a problem in the initialization or management of this USB-C controller in the Linux code base. This is the second time this issue has been reported (and third reportee -- or was the entry I saw in Reddit from one of you two?) and all reports are against Linux (ok, not a big sample set, I admit -- but no one is reproducing this occurring on Windows (are they?)).

...S

[AlHill]

Actually, I did notice that.

Doc

[User1589201909612646]

Hi there, just compiled https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git

I have installed it: Linux 5.7.0-rc5-1-mainline #1 SMP PREEMPT Tue, 12 May 2020 12:40:11 +0000 x86_64 GNU/Linux

After this the issue is still present. ( irq/65-i2c-INT3 )

BTW isn't INT3 a debugging interface for devs?

Thanks, qi

PS sorry about the number username, I already sent Intel a notification about that;)

[User1589201909612646]

@ AlbertR_Intel

The box behaves okay besides the irq beating the box, this constant running (AVG 1.00+) should also lead to higher power consumption, and this box is on 24/7....

The files you requested for to MKonr1. but I think mine are also welcome;)

I also added the interrupts.

Thanks, qi

[User1589201909612646]

I see one can only add one attachment per post:(

Here's the SSU report

[User1589201909612646]

And the dmesg ...

All the files are for ArchLinux running mainline kernel

[Alberto_R_Intel]

User15892019096126469052, Thank you very much for sharing that information.

At this point, we will wait for MKonr1 to provide the information requested previously in order to further assist him with this subject as well.

Regards,

Albert R.

Intel Customer Support Technician

A Contingent Worker at Intel

[MKonr1]

I found temporary solution:

modprobe -r tps6598x

It disables problematic device driver and interrupts from it.

I've contacted with tps6598x driver creator from Intel and he said:

"My guess is that the IRQ resource is not correct for the PD

controller causing you to see irq flood.

The problem is that the ACPI device entry (the node) on this platform

has 4 I2CSerialBus resources and 4 IRQ resources. The idea is that the

single ACPI device entry can represent up to 4 USB PD controllers. The

problem is that there is no way to know which IRQ resource belongs to

which I2CSerialBus resource :-(.

Andy, this is one of those multi-instantiate I2C slave devices with

HID INT3515.

The only solution I can think of is that we start maintaining DMI

quirk table in drivers/platform/x86/i2c-multi-instantiate.c where we

supply the correct i2c to irq resource mapping for every platform

that has this device(s)."

[User1589201909612646]

@ MKonr1

Thank you for figuring out what the problem is.

I would like to know, does it have side effects disabling this tps6598x module and how do I keep track of it's development so I know when to enable it again?

Thanks, qi

edit: NVM, I found the affected file on GitHub.

I will keep a watch on it.

Taking a closer look at the interrupts I found it too.

Disregard my assumptions from post#6.

I'm not a coder, but trying to understand the issue;)

[n_scott_pearson]

What an ugly issue! Why is it not for Windows? (Or is it?)

...S

[User1589201909612646]

@ n.scott.pearson

Probably not, if I understand correctly it's a I2C multi-instantiate driver 'bug' in the Linux kernel.

(see) https://github.com/torvalds/linux/blob/master/drivers/platform/x86/i2c-multi-instantiate.c

qi

[MKonr1]

Disabling this driver affects smth related with USD power delivery feature I think.

About windows, one of engineer from USB driver mailing list said:

"I think they do silly things there in usual obscure MS way, i.e.

hardcoding everything in the driver per platform.

That's why I'm really disappointed how things are going on."

LOL :)

[Alberto_R_Intel]

MKonr1, ma v, Thank you very much for sharing those comments.

We will do further research on this subject to verify if we can provide additional details on this regard. As soon as I get any information I will post all the updates on this thread.

Regards,

Albert R.

Intel Customer Support Technician

A Contingent Worker at Intel

[Ronny_G_Intel]

Hi everyone,

First of all, thank you for the detailed report on this issue and for sharing your findings.

I am bringing this issue up to the escalation team and I will report back to this community as soon as I get any update but I can't promise anything at this point and please keep in mind that we have not validated Linux* with Frost Canyon, see the Operating System validation list here https://www.intel.com/content/dam/support/us/en/documents/intel-nuc/Supported-OS-NUC.pdf, in addition to that, based on your reports this issue is more on the Linux side than in our side but let me check and I will come back to you.

Regards,

Ronny G

[User1589201909612646]

Thanks for you're answers MKonr1 AlbertR & Ronny,

@ Ronny_Intel

Many Linux users use NUC's for they mostly run very well on it.

I wish Intel was/is (from my perspective and I think that of many Linux users) more (than now) part of the Linux community too ;)

It would be great if all NUC's support Linux OOTB even though most of them already run fine on it...

Though, I'm very happy you're willing too look into the issue and maybe even solve it, thanks in advance for that 👍

Thanks, ma v (qi)

PS can you help me change that ugly numbered username into ma v, I did send 'you' messages about that too, thanks...

[Ronny_G_Intel]

Hi ma v,

Unfortunately, after doing some digging I don't have a lot to say about this request. The TPS65982 is a USB Type-C and USB PD Controller built-in the system and it is from Texas Instruments*. The only recommendation I can make at this point is for you to contact their support and work with them to get this device working properly on Linux*, it may require drives or configuration, I am not aware of that. Here is the device information we located on the web: https://www.ti.com/product/TPS65982

In regards to Linux support, we do offer support for validated systems and we try our best for non-validated but depending on the issue this could be very challenging, I hope that you understand this.

On the other side, we are migrating our community over to a new platform so I would wait a couple of weeks, and please try changing your username.

Regards,

Ronny G

[knapoc]

Is there new information available regarding this issue (mailing list threads etc)? I’m experiencing the same on Ubuntu 20.04 LTS.

[inter-links]

Hi all, I have the same problem on NUC10i7, i disabled as advised above, load average is now how it should be.

My other problem is that I could only install Debian 11 (Testing), Ubuntu 20.04 hangs immediately after trying to install. Ubuntu 20.10 installs fine.

But I'd like to run Ubuntu 20.04 Server on my NUC10i7, has anybody made work? I updated the NUC Bios to the latest.

Thank you