Bug 216904

Summary: backlight brighness control with video missing /sys/class/backlight interface
Product: ACPI Reporter: walter59 (walter.moeller)
Component: ACPICA-CoreAssignee: acpi_acpica-core (acpi_acpica-core)
Status: NEW ---    
Severity: high CC: jwrdegoede, kernel, regressions
Priority: P1    
Hardware: Intel   
OS: Linux   
Kernel Version: 6.2-rcx Subsystem:
Regression: No Bisected commit-id:
Attachments: lscpi -nn
dmidecode

Description walter59 2023-01-09 11:31:18 UTC
backlight brighness control with video missing /sys/class/backlight interface

kernel 6.1 it is ok --- kernel 6.2 not
Comment 1 The Linux kernel's regression tracker (Thorsten Leemhuis) 2023-01-10 06:24:45 UTC
What kind of machine is this? Could you please attach the output of `lspci -nn`  and `dmidecode`
Comment 2 Adrien 2023-01-11 21:47:24 UTC
Hi,

Same here. Everything is fine on 6.1.3 and the issue seems to happen starting 6.1.4. 
`ls /sys/class/backlight/` returns nothing.
I can't adjust the brightness with Fn+F3/F4 (dmesg returns acer_wmi: Unknown function number - 4 - 0).
I'll try to do a bisect.
Comment 3 Adrien 2023-01-11 21:48:09 UTC
Created attachment 303582 [details]
lscpi -nn
Comment 4 Adrien 2023-01-11 21:48:55 UTC
Created attachment 303583 [details]
dmidecode
Comment 5 Hans de Goede 2023-01-12 15:33:41 UTC
Thank you for the bug report and sorry about the regression.

What is the output of "ls /sys/class/backlight" with a working kernel like 6.1.3 ?

Also what is the output of "cat /proc/cmdline" ?

Are you perhaps specifying an "acpi_backlight=xxx" argument on the kernel commandline? If you are specifying this, try dropping this from the kernel commandline that will likely fix things.
Comment 6 Adrien 2023-01-12 15:58:39 UTC
Thank you for your response.

With 6.1.3 "ls /sys/class/backlight" returns amdgpu_bl0. 

Output of "cat /proc/cmdline" returns 
BOOT_IMAGE=/boot/vmlinuz-linux root=UUID=7d604fa9-81a1-4cbc-ae1d-8brfe544148b rw quiet loglevel=3 splash vt.global_cursor_default=0 resume=UUID=e40cbf4e-5477-4ecb-b1f1-bd8d5890f507 audit=0 nvme_load=yes sysrq_always_enabled=1 mem_sleep_default=deep
in 6.1.4 **after** I deleted the "acpi_backlight=vendor" I had. 

Thank you, dropping acpi_backlight=vendor do bring back amdgpu_bl0 to /sys/class/backlight. 
However, dmesg now shows "acer_wmi: Unknown function number - 4 - 0" every time the brightness change key are touched, but it does not seem to have any effect (brightness do change) but to spam the log.

I don't know if the issue is really related to walter59.
Comment 7 Hans de Goede 2023-01-12 16:15:26 UTC
Adrien, Thank you for testing without acpi_backlight=vendor. 6.1.4 got a fix from 6.2 backported which makes the driver providing amdgpu_bl0 honor acpi_backlight=vendor which is what makes it goes away.

The amdgpu backlight counts as native, so to manually select that you would pass acpi_backlight=native. But it is best to just let the autodetect code / heuristics do their thing.

> However, dmesg now shows "acer_wmi: Unknown function number - 4 - 0" every
> time the brightness change key are touched, but it does not seem to have any
> effect (brightness do change) but to spam the log.

Did you check to see if those messages are there with 6.1.3 too?

My first guess would be that those messages have been there all along, but you just never noticed them...
Comment 8 Artem S. Tashkinov 2023-01-22 11:30:28 UTC
*** Bug 216947 has been marked as a duplicate of this bug. ***