search

hexdump0815 / linux-mainline-mediatek-mt81xx-kernel

linux-mainline-mediatek-mt81xx-kernel
15 stars 7 forks source link

Problems with LTS and Linux Stable #12

Open nvidialinuxuser opened 9 months ago

nvidialinuxuser commented 9 months ago

Hi,

Just here to give a quick warning that there's some odd regression in Linux stable that appears to break USB booting from my testing on Kukui. This issue is not specific to any filesystem or Linux distribution. I was able to confirm this regression with the Linux stable 6.6.7 release. Booting fails entirely with a load of "Buffer I/O error" messages as well as "lost sync page write" but I do recall USB booting working properly with Linux 6.1 LTS though. It looks like on Linux stable is it either corrupting and/or failing to read contents correctly. I should clarify that I used your patches with Linux stable.

hexdump0815 commented 9 months ago

thanks a lot for this info - i so far did some very basic testing of v6.6 on my kukui kappa on emmc and there it worked, but hat its own set of regressions (see: https://github.com/hexdump0815/linux-mainline-mediatek-mt81xx-kernel/commit/6a3ca0b79dc357fbad142596ba65c3edc2c530ed ) ... i did not have time and energy to look deeper yet

nvidialinuxuser commented 9 months ago

If it helps, this was tested on a Fennel 14 SKU2 ChromeBook. I did try two USB drives that I know for a fact are not problematic. I should note that one of your USB patches broke on the latest mainline RC. I tried to investigate it. it looks like they've changed a few things so your patch doesn't apply correctly. I wasn't able to successfully drop in a patched version of the older driver as the build would fail.

hexdump0815 commented 9 months ago

i have dropped two of the usb-patches as they seemed to be included now - maybe have a look at the branch my above link points to - this is from my v6.6 try ... i hope to find some time to look a bit more into v6.6 between the years

nvidialinuxuser commented 9 months ago

I'm not exactly sure how I would be able to debug this issue. It definitely seems like something upstream broke it. Bitrot already? At one point in the past, I would like to think around kernel 6.4 or 6.5 did touchpad support quit working on this device.

nvidialinuxuser commented 9 months ago

You have some patches that no longer work with Linux 6.1 LTS. Looks like things have changed with Linux 6.1 LTS.

Patch File: mt8183-cadmium-kukui.enable-ite.patch Hunk 3 fails at 507. Patch File: mt8183-kukui-disable-unsupported-thermal-zones.patch Hunk 1 fails at 739. Hunk 2 fails at 791.

hexdump0815 commented 8 months ago

@nvidialinuxuser - that happens from time to time due to backported changes/fixes - maybe give v6.6 a try (currently in the master branch here - i'll create a lts-v6.6 branch from it soon) ... v6.6 built from it seems to run quite well as i'm typing this right now on a kappa running v6.6.9 :)

nvidialinuxuser commented 8 months ago

@hexdump0815, I got a different result this time (I just built the latest 6.6 stable not very long ago with the needed patches), rather than it being hung on buffer I/O errors, it's now being stuck on some kind of process related to the UUID if of my Linux root. I'm using OpenSuse. Nothing seems to start incorrectly. After doing some research, this could potentially be related to those USB errors that I've been having. I actually had trouble earlier where it would get stuck at somewhere else related to the filesystem I was using, but I managed to resolve that.

nvidialinuxuser commented 8 months ago

Tried Fedora as well. Got a very similar issue as I did with OpenSuse initially which I also worked around it. With Fedora, I just simply had removed my initramfs file from the 'kernel.its' file. This time, it nearly booted successfully, except there's a ton of I/O errors, write failures, and "Input/output" error messages. I'm guessing that it correctly reads USB drives but it fails to write USB drives.