[prev in list] [next in list] [prev in thread] [next in thread] 

List:       linuxbios
Subject:    [coreboot] Re: Please test coreboot using laptops with kernel 6.1-rc# for a possible backlight regre
From:       Hans de Goede <hdegoede () redhat ! com>
Date:       2022-10-27 10:56:41
Message-ID: 3d589732-d3f1-eeb6-d7c4-1f812807015a () redhat ! com
[Download RAW message or body]

Hi,

On 10/26/22 14:03, Hans de Goede wrote:
> Hi All,
> 
> I have landed a large(ish) refactor of the ACPI/x86 backlight detection
> code in the kernel for 6.1. It has come to my attention that this may
> cause backlight control to stop working on laptops where the original
> BIOS has been replaced with coreboot, see this patch, which is the root
> cause of the possible regression + its replies:
> https://lore.kernel.org/platform-driver-x86/20220825143726.269890-3-hdegoede@redhat.com/

Note I have written a blogpost to ask for more general testing of
old/weird laptops for this. This includes an easy way to identify if
your laptop might be affected (without needing to build a 6.1-rc#
kernel). This easy way should also apply to coreboot using laptops:

https://hansdegoede.dreamwidth.org/26548.html

Please give this a try if possible.

Regards,

Hans



> On ACPI/x86 laptops there may be multiple methods available to control
> the backlight (native GPU control, ACPI video bus, vendor specific)
> and even though a method is available it does not always work, basically
> the only method guaranteed to work is the one preferred by the Windows
> version with which the laptop shipped.
> 
> The kernel so far has delegated part of this problem by registering
> multiple backlight devices under /sys/class/backlight and then
> letting userspace figure it out. One goal of the refactor is to
> only register 1 backlight device per panel.
> 
> On many (somewhat older) laptops "ls /sys/class/backlight" would
> show 2 backlight devices: "acpi_video0" and "intel_backlight"
> and the various (per desktop) userspace backlight helpers would
> prefer the "acpi_video0". The goal of the linked kernel patch
> is to simply not register "intel_backlight" in this case, just
> like the kernel was already not registering "acpi_video0" on the
> models where "intel_backlight" should be used.
> 
> But this may cause problems / a change in behavior if:
> 
> 1. You have an unusual / or customized userspace which prefers
> intel_backlight over acpi_video0 when both are present
> 
> 2. You always only had intel_backlight and that now no longer
> gets registered, leaving you completely without backlight support.
> 
> I would like to try and find and resolve any such problems before
> 6.1 gets released. As such if you are using a laptop with coreboot
> please run the following tests:
> 
> 
> 1. Boot kernel 6.0 or older, and collect the output of:
> 
> ls /sys/class/backlight
> 
> 
> 2. Boot kernel 6.1, note for ChromeBooks please used Torvalds
>    latest master and not rc2 since a fix for ChromeBooks has just
>    landed, and collect the output of:
> 
> ls /sys/class/backlight
> 
> 
> 3. Check if everything still works as expected with kernel 6.1
> 
> Report the results of 1-3 either on the list or in a direct
> email to me (your choice).
> 
> Please include the model of your laptop and please report this
> even if everything still works fine, I need to know how things
> work on coreboot laptops which have not regressed too, to avoid
> accidentally regressing those if I need to do fixes for some
> other coreboot capable models.
> 
> 
> 4. Run:
> 
> sudo dmidecode > dmidecode.txt
> sudo acpidump -o acpidump.txt
> 
> And then send me a *private* off-list email with the
> dmidecode.txt and acpidump.txt files attached. Please don't
> forget to mention the model of your laptop in this email.
> 
> 
> Sorry for any breakage my backlight refactor may have
> caused and thank you in advance for taking the time to
> test this.
> 
> Regards,
> 
> Hans
> 

_______________________________________________
coreboot mailing list -- coreboot@coreboot.org
To unsubscribe send an email to coreboot-leave@coreboot.org
[prev in list] [next in list] [prev in thread] [next in thread]