Scheduled Maintenance: We are aware of an issue with Google, AOL, and Yahoo services as email providers which are blocking new registrations. We are trying to fix the issue and we have several internal and external support tickets in process to resolve the issue. Please see: viewtopic.php?t=158230

 

 

 

Post-5.10 kernels bring new laptop backlight issues

Need help with peripherals or devices?
Post Reply
Message
Author
Fluora
Posts: 4
Joined: 2021-11-30 15:36

Post-5.10 kernels bring new laptop backlight issues

#1 Post by Fluora »

I'm using a Panasonic Toughbook CF-54, containing an Intel Core i5-5300U and a display connected via eDP. I'm running Debian Testing (Bookworm) at the moment, which is on kernel version 5.15.

On kernel versions 5.10 and prior, the display brightness was not adjustable with the default settings, though the brightness keys were read by the OS and could "pretend" to adjust the brightness in software, though without actually affecting the display backlight. I added "i915.enable_dpcd_backlight=1 acpi_osi=" to my kernel parameters to fix this, which resulted in expected behavior (i.e. same as before, except the backlight actually changes when the OS tells it to).

Kernels after 5.10 - namely 5.13 and 5.15, though it may affect kernels as early as 5.11 - have worsened the situation. On 5.15, "i915.enable_dpcd_backlight" has no apparent effect at all, and everything depends on "acpi_osi". If it's set to the empty string ("acpi_osi=") then the backlight can be adjusted by the OS, but the brightness keys do not function at all and are never detected by the OS in any way. If I set "acpi_osi" to "Windows", "Linux", or "!Windows 2012", or do not set it at all, then the display brightness keys work and are detected by the OS, but the display backlight itself cannot be adjusted at all and does not respond. Basically, I can have either of the things I need individually, but never both at once.

Things I have also tried:
- Setting "acpi_backlight" to "vendor" or "native". When combined with "acpi_osi=" on any kernel version, "vendor" makes the backlight setting resolution better and "native" makes it much worse, but has no other effect.
- Setting "pcie_aspm=force", at the suggestion of an Arch forum post. No detectable effect.

As a tangential note, I also experienced a regression in Bluetooth at exactly the same time as the display backlight issues, wherein the Bluetooth part of my Intel wireless card freezes on every boot and doesn't respond until I turn the physical wireless switch off and on again. I don't have any reason to think this is related, but I'm mentioning it anyway since it did start at the same time as this other new problem.

Anyway: can someone help me understand what "acpi_osi=" is actually doing? I have heard that Linux's default behavior is to "lie" to the firmware and say that Linux is not running when the firmware asks, since many devices disable important features if they think an OS other than Windows is running, and also that this behavior can be changed to make it "tell the truth" by setting "acpi_osi=Linux". However, I don't think I understand what the effect of setting "acpi_osi" to the empty string is. It clearly has some effect on whether or not my display brightness keys and backlight control work, but not a consistent effect. What's going on here?

I've seen some forum posts here and there that have a very similar problem and seem to believe that it's related to their Nvidia graphics cards, but my machine does not have an Nvidia card (or any dedicated video card at all). Is this problem likely to be graphics-related, or is it some other thing to do with ACPI?

Anyway, thanks in advance for any insight anyone can offer, and let me know if you have anything you want me to try and report back on.

To be clear, I'm looking for a correct general solution to this problem that will work for others who might encounter it - either a kernel parameter setting that makes it work correctly, or a way to gather enough information to submit a kernel bug report. I appreciate the utility of certain "hacks" I've seen, like redirecting certain sysfs interfaces to others using custom scripts, but I'd like to avoid that kind of crude temporary solution - at least until I have at least some sense that a more serious solution is on the way.

In the meantime, I can just use the Stable kernel version (5.10) and do okay, but I'm of course not going to be able to keep that forever, so I'd very much like to find a way to get my system to work with newer kernels soon.

LE_746F6D617A7A69
Posts: 932
Joined: 2020-05-03 14:16
Has thanked: 7 times
Been thanked: 65 times

Re: Post-5.10 kernels bring new laptop backlight issues

#2 Post by LE_746F6D617A7A69 »

Fluora wrote: 2021-11-30 17:13 Anyway: can someone help me understand what "acpi_osi=" is actually doing?
Official documentation:
https://www.kernel.org/doc/html/latest/ ... i/osi.html
The above documentation probably won't help You to solve the problem - but at least it should help to understand where is the problem.

In general, You should try to upgrade the BIOS - all the Panasonic hardened laptops have broken BIOSes and the upgrade *might* solve the problem.

Alternatively, You may need/want to add a specific "quirk" to the ACPI kernel code, and compile the kernel for Your specific HW, but anyway I would suggest to report a BUG.

BTW: Do You have any ACPI errors in the logs? (like $>dmesg | grep ACPI ?)
Bill Gates: "(...) In my case, I went to the garbage cans at the Computer Science Center and I fished out listings of their operating system."
The_full_story and Nothing_have_changed

Fluora
Posts: 4
Joined: 2021-11-30 15:36

Re: Post-5.10 kernels bring new laptop backlight issues

#3 Post by Fluora »

LE_746F6D617A7A69 wrote: 2021-11-30 22:48 In general, You should try to upgrade the BIOS - all the Panasonic hardened laptops have broken BIOSes and the upgrade *might* solve the problem.
Can you elaborate on what you mean by this? Are you referring to the Management Engine-related security vulnerabilities, or something else more specific to Toughbooks?
Either way, I haven't run Windows on anything in many years and a firmware update doesn't sound like much fun at all. But if that's what's necessary, so be it, I guess.
BTW: Do You have any ACPI errors in the logs? (like $>dmesg | grep ACPI ?)
I'll get back to you on this once I have a chance to boot both kernel versions and have a look at the dmesg output produced by each one. Tomorrow, at the latest.

Fluora
Posts: 4
Joined: 2021-11-30 15:36

Re: Post-5.10 kernels bring new laptop backlight issues

#4 Post by Fluora »

Okay, I'm a little late. Sorry about that. I spent a while going through the dmesg output, but I'm not really sure what I'm looking for. There's a lot of stuff related to ACPI, and a lot of things that are different between kernel versions, but I have yet to spot an obvious error message.

Here's the full dmesg output for the two versions.
5.10: https://pastebin.com/XDCScyTu
5.15: https://pastebin.com/hF5rJuQe

If you'd like me to condense those down to just the ACPI messages, I can do that, but I figured it made more sense to just post the whole nine yards - it's easy to grep it after the fact, after all.

I'm still curious to hear about the Toughbook firmware situation you alluded to, by the way.

Fluora
Posts: 4
Joined: 2021-11-30 15:36

Re: Post-5.10 kernels bring new laptop backlight issues

#5 Post by Fluora »

LE_746F6D617A7A69 wrote: 2021-11-30 22:48
Fluora wrote: 2021-11-30 17:13 Anyway: can someone help me understand what "acpi_osi=" is actually doing?
Official documentation:
https://www.kernel.org/doc/html/latest/ ... i/osi.html
The above documentation probably won't help You to solve the problem - but at least it should help to understand where is the problem.

In general, You should try to upgrade the BIOS - all the Panasonic hardened laptops have broken BIOSes and the upgrade *might* solve the problem.

Alternatively, You may need/want to add a specific "quirk" to the ACPI kernel code, and compile the kernel for Your specific HW, but anyway I would suggest to report a BUG.

BTW: Do You have any ACPI errors in the logs? (like $>dmesg | grep ACPI ?)
I have updated the BIOS and EC firmware to the latest versions from Panasonic, and I see no change in behavior at all. Apart from version number increments, I can't tell that anything has changed. It's all exactly the same as it was before.

I'm okay with falling back on the 5.10 kernel for now, since that's what Stable is on anyway and will be for a while, but my goal here is to see my laptop working with later kernels by the time Bookworm stabilizes. That's a long way off, but I get the feeling that debugging this is going to take a long time, so I assume getting started now is the right idea.

Do you have any pointers on where to begin on this? I've never submitted a kernel bug report before, so I wouldn't mind a quick overview of what type of info is expected to be included in those, if you should happen to know.

Post Reply