kernel_task constantly at ~70%, fans running at max speed constantly #17
Comments
|
Hey, so far the issue has not been resolved and I still cannot not figure out why it happens. I believe this must be either fixed at an OpenCore level or needs a better SSDT patch then what I have attempted. |
|
Thanks for your reply. Aside from the high CPU usage that can be observed in the activity monitor, how were you able to confirm that interrupt gpe6f was the root cause? I was unable to find any relevant log information anywhere during my attempts. |
|
Does Big Sur EFI have this same issue ? |
|
@sambow23 I came across your post on https://osxlatitude.com/forums/topic/17098-dell-xps-7390-high-cpu-usagehorrible-battery-life/ while digging into this issue, have you managed to resolve it with this method ? |
As stated in the readme, the broken interrupt has the exact same behavior (100% CPU on one or two cores, fan constantly going off) on Linux when its not disabled. if you use a power draw measuring tool on linux such as |
Any EFI exhibits this issue.
At the time I believed the issue was solved because my fans stopped running at 100% all the time. But that could've also been fixed when I was changing SMBIOS's until I found one with a lower TDP limit. On macOS its very difficult to know if the problematic interrupt was ever disabled as there is no userspace utility to check as there is on Linux. |
|
maybe you can try this EFI https://github.com/H3C4T0M8/Hackintosh-Dell-XPS-13-7390-OpenCore i dont have any 100% fans, but audio jack broken |
i will try this tomorrow! thanks! |
|
@irfanhutomo2 Thanks for the link. I have just tested your EFI, but I believe my screen specs (4K) differs from the intended one, causing a black screen to be shown after boot. I can hear the bell sound from macOS when pressing keys, which leads me to believe that the OS is booted, but nothing is showing on the screen. Would you mind sharing your screen specifications on this model? |
|
@sintaxx @irfanhutomo2 I have tried the EFI 0.8.4 on the link that was provided. As per my previous message: the built-in screen is totally black. However, connecting an external screen proved to be working. That is how I could confirm that the system is booting correctly. To the bad news now: kernel_task is still using a high amount of %.
And fans are still roaring. I tried using the SSDT-THUNDERBOLT.aml from the repository, since it seems to implement a lot more logic than then SSDT-DTB3.aml provided in this repository. Unfortunately, kernel_task still uses a fairly high amount of CPU % (around 60%). I would love to hear more feedback around this one. If anyone else gives it a go. |
|
Going a bit further: I wanted to double check the hypothesis behind gpe6F ACPI interrupt being the actual cause of this high kernel_task usage. I believe it might have been a valid one a few months ago (?), but I just tested a fresh Ubuntu install, using the laptop updated with the latest BIOS available (1.17.0, released on the 9th of August). Now, even with /sys/firmware/acpi/interrupts/gpe6F left untouched and marked as enable, CPU usage remains steady at < 1% when nothing is running. Thunderbolt enabled or disabled in the BIOS, behavior is consistent. Back to macOS now, running version 12.5.1, the kernel_task still is gobbling up CPU unfortunately. Just mentioning this as it was assumed that this was the problem causing the high kernel_task CPU % usage, but it might be something else entirely ... @irfanhutomo2 Can you confirm that the EFI you suggested applies to this exact model of Dell laptop? I know there is another XPS 13" 7390 model, called "2 in 1". |
Good to know it's been fixed under Linux, either that being from Linux kernel updates or the BIOS. I don't currently have my 7390 to update the BIOS myself and try macOS but I should be able to get it soon. |
my bios is latest and it isn't resolved in big sur |
It might of course have been fixed in a more recent kernel version, but I tend to believe it is fixed in the latest BIOS (1.17.0). The reason is that I tested this with the USB install key I still had, of an Ubuntu 22.04 beta from late March 2022: in the usb live trial environment, the problem could be observed then, with the kernel shipped on this beta version, but not this time. |
|
Moving further, I have now successfully confirmed it is thunderbolt related. After enabling all debug for ACPI, following the guide and recommendations for debugging, I get the following kernel logs Kernel logs filtered on ACPI activity only: All kernel logs: There is one entry that stands out, as it is literally spamming the entire stream:
In the span of 1 second, this log entry appears in average 50 times. So clearly, this should be the cause. According to the log formatting, this should be linked to ThunderboltReset.kext … I attach the full log files for future reference to this message. In case someone has more time than me to dig deeper into this 😃 |
|
fyi, i disable thunderbolt on bios and fans not running at max speed constantly |
Hi,
i have been trying out your experimental EFI for Monterey, updated everything to the latest available version, updated the BIOS of my machine to 1.16.0 (released earlier this month). Unfortunately, it very much still behaves the same: power is being drained super quickly and fans are running at max speed constantly. Only notable thing is : kernel_task is running at 70% CPU usage constantly.
I looked at what you had done and it should do the trick. I was wondering if you did succeed at making the hack perfect ? Apart from this issue, I must admit that it is a brilliant machine and all works super well.
FYI: I tried asking for some guidance and support over the open core Discord channels, but did not get any answer. I just posted this message in Tonymacx86 forum: https://www.tonymacx86.com/threads/kernel_task-using-high-of-cpu-probably-caused-by-gpe-acpi-interrupt.321578/
Thanks in advance,
Eric
The text was updated successfully, but these errors were encountered: