Closed andersfugmann closed 2 years ago
I’d rather want to have a review by @Redecorating and @aunali1 on this.
Btw. This kernel error log messages was observed on v5.17-1
$ uname -a
Linux trillian 5.17.0-t2 #1 SMP PREEMPT Wed Mar 23 12:03:25 UTC 2022 x86_64 GNU/Linux
Oh I see. So not in 5.16.x or 5.15.x?
Also, have you tested this patch?
Actually not. I was hoping that GH Actions could test if it compiles. Let me see if I can create the packages locally and create a proper PR.
Well then you could just add the patch to the patches folder and create a PR for the same.
That's exactly what I was planning to (and did) :-)
See #15
Alright. Lemme test it on my side and then send it to the collection of other patches.
Well, this bug seems to affect the older 5.16.x and 5.15.x kernels as well. Well, I am accepting the patch and backporting it as well. You probably could make the developers aware of the fact that it needs to be backported to stable as well.
New kernels with your patch shall be released soon.
Awesome - Thanks. The ACPICA developers are well aware that the bug existed in at least Linux 5.15 (As they reference kernel 5.15rc3 in the commit message for the fix). I don't know why I did not see this error before, but errors are indeed also present in my kernel logs from 5.16.
UBSAN complains about using array index of
-1
when resolving ACPI operands when there are none.A more detailed description of the bug + possible solution can be seen here: https://github.com/acpica/acpica/pull/745 The bug seems to be triggered only on apple MBP's, but the solution seems correct for all architectures.
Could we add a patch to add a patch for this to the T2 kernels until the fix finds its way into the linux kernel:
patches/1003-fix-acpica-for-zero-arguments-acpi-calls.patch
: