https://bugzilla.kernel.org/show_bug.cgi?id=221241
Bug ID: 221241
Summary: slab object leak during "ACPI BIOS Error (bug): Could
not resolve symbol ..." path
Product: ACPI
Version: 2.5
Hardware: Intel
OS: Linux
Status: NEW
Severity: low
Priority: P3
Component: ACPICA-Core
Assignee: [email protected]
Reporter: [email protected]
Regression: No
Created attachment 309671
--> https://bugzilla.kernel.org/attachment.cgi?id=309671&action=edit
kernel log with debug output enabled
Observed in 6.19.6-arch1-1 kernel, which I confirmed has no patches to the
drivers/acpi/ tree.
Upon waking from sleep, the kernel logs are spammed with:
> Mar 14 10:29:58 persephone kernel: ACPI BIOS Error (bug): Could not resolve
> symbol [\_GPE.AL6B.WAK0], AE_NOT_FOUND (20250807/psargs-332)
> Mar 14 10:29:58 persephone kernel: ACPI Error: Aborting method \_GPE.AL6B due
> to previous error (AE_NOT_FOUND) (20250807/psparse-529)
> Mar 14 10:29:58 persephone kernel: ACPI Error: Aborting method \_GPE._L6B due
> to previous error (AE_NOT_FOUND) (20250807/psparse-529)
> Mar 14 10:29:58 persephone kernel: ACPI Error: AE_NOT_FOUND, while evaluating
> GPE method [_L6B] (20250807/evgpe-511)
This happens hundreds or thousands of times a second. This bug report is not
about the specific not found device(?), but rather that as this error occurs,
slab objects are allocated and not released, leading to a steady leak in
memory. Within a number of hours this system is exhausted of it's 32G of memory
and panics.
For example, `slabtop` reports the Acpi-State cache as having 76,026,465
objects, all of which are active after ~12 hours of runtime following the first
sleep.
I managed to enable debug tracing and debug level via this script and I've
attached the results. I hope they are helpful.
> startat=$(date +"%H:%M:%S.%N")
> echo "0xFFFFFFFF" > /sys/module/acpi/parameters/debug_layer
> echo "0xFFFFFFFF" > /sys/module/acpi/parameters/debug_level
> sleep 0.2
> echo "0x0000000C" > /sys/module/acpi/parameters/debug_level
> echo "0x00000000" > /sys/module/acpi/parameters/debug_layer
> endat=$(date +"%H:%M:%S.%N")
> journalctl -k --since "$startat" --until "$endat" > log.txt
I'm happy to run additional reproduction or debug gathering on this system if
needed.
--
You may reply to this email to add a comment.
You are receiving this mail because:
You are watching the assignee of the bug.
_______________________________________________
acpi-bugzilla mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/acpi-bugzilla
