Re: [PATCH] ACPI: video: Use vendor backlight on Lenovo X1 Carbon.

[dengxiang <dengxiang@nfschina.com>]

Hi Hans,

> A couple of remarks / questions:

> 1. Looking at the strings you match on this is not for a Lenovo X1 Carbon,
> but rather for a Lenovo Kaitan model ?  So it seems that the commit message
> and the comment for the quirk need some work.

ok, I will add DMI_PRODUCT_VERSION & DMI_BOARD_NAME to make a distinction between  X1 Carbon and other kaitian models.

> 2. I have never heard of a zx_backlight interface before and there certainly
> is no upstream driver providing this. I believe you need to explain what
> is going on in a bit more detail here and then we can see if this really is
> the best way to fix this. It seems that these Lenovo Kaitan laptops are
> using Zhaoxin Kaixian x86 processors with integrate graphics. I would expect
> the zx_backlight interface to be provided by the driver for the Zhaoxin Kaixian
> integrated graphics in this case. And if that is the case then the integrated
> graphics driver should use BACKLIGHT_RAW (aka native) for the backlight type
> and with that change this quirk should not be necessary .

Yes, zx_backlight interface has been provided by the driver for the Zhaoxin Kaixian integrated graphics. Also use backlight_device_register("zx_backlight",...).
Strangely enough, X1 Carbon laptops will generate two native acpi_video as belows:
 
lrwxrwxrwx 1 root root 0  5月 14 16:20 acpi_video0 -> ../../devices/pci0000:00/0000:00:01.0/backlight/acpi_video0
lrwxrwxrwx 1 root root 0  5月 14 16:20 acpi_video1 -> ../../devices/pci0000:00/0000:00:01.0/backlight/acpi_video1

As you see, it will conflict with the same pci bus, then zx_blacklight interface can't be shown on the path /sys/class/backlight/.
That is to say, zhaoxin driver contain key code as belows:
#if DRM_VERSION_CODE >= KERNEL_VERSION(4, 2, 0)
    if(acpi_video_get_backlight_type() != acpi_backlight_vendor)
    {
        return ret;
    }
#endif

If i remove the key code, this laptops will generate two native acpi_video and zx_backlight on the sys backlight patch. Once add acpi_backlight=vendor parameter into kernel cmdline, 
just zx_backlight interface has been left on the sys path, which mean that both acpi_video0 and acpi_video1 interface can not be found.

> 3. Vendor specific backlight interfaces are normally only found on really
> old laptops. Since Windows XP laptops typically use the ACPI backlight
> interface and since Windows 8 they typically use the GPU's native
> backlight driver. So adding a quirk to use a vendor interface in 2024 is
> weird. Again can you explain in a lot more detail what is going on here,
> but I guess the backlight class device is provided by the driver for the
> integrated graphics and in that case the fix is to simply change the type
> of the backlight device registered by the igfx driver to BACKLIGHT_RAW.

As mentioned in 2  questions above zhaoxin drivers had used backlight_device_register("zx_backlight"...) as BACKLIGHT_RAW.

> 4. You posted the same patch twice ?

Sorry, i was wrong to think that before patch would be missed by you. also i forgot about the time zone difference. I am sorry for any inconvenience that I have brought to you.

Best Regards,
dengxiang