https://bugzilla.kernel.org/show_bug.cgi?id=221866

--- Comment #19 from David Del Sol ([email protected]) ---
Important update: the slow-charge bug no longer reproduces on the current
kernel, and I can now show it was never OSI-related.

Kernel linux-cachyos was auto-updated 7.1.6 -> 7.1.8 in the middle of my test
series today, which is the key variable:

- 7.1.6, after 30s EC reset, stock cmdline, AC present at power-on: ~0.2-2.5 W
(bug reproduces)
- 7.1.8, same 30s EC reset procedure (identical - the hold also resets the BIOS
in all cases, verified), stock cmdline, AC present at power-on: charges
normally (~20 W, high-SoC taper; battery 85-88%)
- 7.1.8, same reset, workaround cmdline: 28.8 W at 79%

All three 30s holds behaved identically (including the BIOS reset / re-enabling
Secure Boot), so the BIOS state was constant - the only variable that changed
between the 2.5 W and normal-charging results is the kernel release.

Consistent with this, the OSI claim machinery observably changed between the
two kernels (requesting _OSI("Windows 2022") logs an Add on 7.1.8, which it
never did on 7.1.6 - default list and/or logging behavior). Upstream
7.1.6->7.1.8 contains no battery/EC/charging commits visible in the changelog
(only max17040/macsmc/bq25890, none relevant here), so the exact mechanism is
still unidentified - I'll dig into the CachyOS fork patches.

The acpi_osi workaround was a correlation artifact on 7.1.6 (Linux
default-claimed the string; no firmware code checks it). Withdrawing the patch
stands; this report documents a firmware-level slow-charge state that kernels
<=7.1.6 failed to clear and 7.1.8 handles correctly.

-- 
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

Reply via email to