Re: ACPI Video Driver creates backlight on desktop board

Re: ACPI Video Driver creates backlight on desktop board

[Hans de Goede]

Hi,

On 3/29/20 11:36 AM, Paul Menzel wrote:
> Dear Linux folks,
> 
> 
> On the MSI desktop board [1]
> 
>      [    0.000000] DMI: Micro-Star International Co., Ltd. MS-7A37/B350M MORTAR (MS-7A37), BIOS 1.MR 12/02/2019
> 
> with an AMD Ryzen 3 2200G with Radeon Vega Graphics, the ACPI Video Driver `video` is loaded and creates a backlight device.
> 
>      $ readlink -f /sys/class/backlight/acpi_video0
>      /sys/devices/pci0000:00/0000:00:08.1/0000:26:00.0/backlight/acpi_video0
> 
> I wonder what the driver is used for as the AMDGPU driver exists for the graphics device.

Backlight on x86 hw is a bit of a mess, there is an ACPI standard for accessing it
and with older (Windows XP era) laptops that is the interface which usually works,
then there are a bunch of vendor specific SMBIOS or WMI backlight interfaces and
then there is the option of directly accessing the hardware as the amdgpu driver
is doing.

We have a bunch of heuristics to avoid the acpi_video driver registering a
backlight interface when it should not, either because the direct hw access
should be used instead; or because there simply is no builtin LCD panel and thus
no backlight to control.

These heuristics are failing on your board.

> If it’s useful, is there a way to prevent the backlight interface from getting created?

You can pass "acpi_backlight=ignore" on the kernel commandline to disable the
interface, this will stop e.g. gnome from showing a non working brightness
slider in its top right system menu.

If this works you can make this permanent and avoid other users from having to
do the same thing by writing a patch adding a quirk for this like this one:

https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=d21a91629f4b8e794fc4c0e0c17c85cedf1d806c

Regards,

Hans

Re: ACPI Video Driver creates backlight on desktop board

[Randy Dunlap]

On 3/30/20 1:51 PM, Hans de Goede wrote:
> Hi,
> 
> On 3/29/20 11:36 AM, Paul Menzel wrote:
>> Dear Linux folks,
>>
>>
>> On the MSI desktop board [1]
>>
>>      [    0.000000] DMI: Micro-Star International Co., Ltd. MS-7A37/B350M MORTAR (MS-7A37), BIOS 1.MR 12/02/2019
>>
>> with an AMD Ryzen 3 2200G with Radeon Vega Graphics, the ACPI Video Driver `video` is loaded and creates a backlight device.
>>
>>      $ readlink -f /sys/class/backlight/acpi_video0
>>      /sys/devices/pci0000:00/0000:00:08.1/0000:26:00.0/backlight/acpi_video0
>>
>> I wonder what the driver is used for as the AMDGPU driver exists for the graphics device.
> 
> Backlight on x86 hw is a bit of a mess, there is an ACPI standard for accessing it
> and with older (Windows XP era) laptops that is the interface which usually works,
> then there are a bunch of vendor specific SMBIOS or WMI backlight interfaces and
> then there is the option of directly accessing the hardware as the amdgpu driver
> is doing.
> 
> We have a bunch of heuristics to avoid the acpi_video driver registering a
> backlight interface when it should not, either because the direct hw access
> should be used instead; or because there simply is no builtin LCD panel and thus
> no backlight to control.
> 
> These heuristics are failing on your board.
> 
>> If it’s useful, is there a way to prevent the backlight interface from getting created?
> 
> You can pass "acpi_backlight=ignore" on the kernel commandline to disable the

Hi Hans,
Should that be "acpi_backlight=none"?  I don't see 'ignore' allowed here:

static void acpi_video_parse_cmdline(void)
{
	if (!strcmp("vendor", acpi_video_backlight_string))
		acpi_backlight_cmdline = acpi_backlight_vendor;
	if (!strcmp("video", acpi_video_backlight_string))
		acpi_backlight_cmdline = acpi_backlight_video;
	if (!strcmp("native", acpi_video_backlight_string))
		acpi_backlight_cmdline = acpi_backlight_native;
	if (!strcmp("none", acpi_video_backlight_string))
		acpi_backlight_cmdline = acpi_backlight_none;
}

and Documentation/admin-guide/kernel-parameters.txt could stand to be updated
with a few of those options.


> interface, this will stop e.g. gnome from showing a non working brightness
> slider in its top right system menu.
> 
> If this works you can make this permanent and avoid other users from having to
> do the same thing by writing a patch adding a quirk for this like this one:
> 
> https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=d21a91629f4b8e794fc4c0e0c17c85cedf1d806c


-- 
~Randy

Re: ACPI Video Driver creates backlight on desktop board

[Hans de Goede]

Hi,

On 3/30/20 11:25 PM, Randy Dunlap wrote:
> On 3/30/20 1:51 PM, Hans de Goede wrote:
>> Hi,
>>
>> On 3/29/20 11:36 AM, Paul Menzel wrote:
>>> Dear Linux folks,
>>>
>>>
>>> On the MSI desktop board [1]
>>>
>>>       [    0.000000] DMI: Micro-Star International Co., Ltd. MS-7A37/B350M MORTAR (MS-7A37), BIOS 1.MR 12/02/2019
>>>
>>> with an AMD Ryzen 3 2200G with Radeon Vega Graphics, the ACPI Video Driver `video` is loaded and creates a backlight device.
>>>
>>>       $ readlink -f /sys/class/backlight/acpi_video0
>>>       /sys/devices/pci0000:00/0000:00:08.1/0000:26:00.0/backlight/acpi_video0
>>>
>>> I wonder what the driver is used for as the AMDGPU driver exists for the graphics device.
>>
>> Backlight on x86 hw is a bit of a mess, there is an ACPI standard for accessing it
>> and with older (Windows XP era) laptops that is the interface which usually works,
>> then there are a bunch of vendor specific SMBIOS or WMI backlight interfaces and
>> then there is the option of directly accessing the hardware as the amdgpu driver
>> is doing.
>>
>> We have a bunch of heuristics to avoid the acpi_video driver registering a
>> backlight interface when it should not, either because the direct hw access
>> should be used instead; or because there simply is no builtin LCD panel and thus
>> no backlight to control.
>>
>> These heuristics are failing on your board.
>>
>>> If it’s useful, is there a way to prevent the backlight interface from getting created?
>>
>> You can pass "acpi_backlight=ignore" on the kernel commandline to disable the
> 
> Hi Hans,
> Should that be "acpi_backlight=none"?  I don't see 'ignore' allowed here:
> 
> static void acpi_video_parse_cmdline(void)
> {
> 	if (!strcmp("vendor", acpi_video_backlight_string))
> 		acpi_backlight_cmdline = acpi_backlight_vendor;
> 	if (!strcmp("video", acpi_video_backlight_string))
> 		acpi_backlight_cmdline = acpi_backlight_video;
> 	if (!strcmp("native", acpi_video_backlight_string))
> 		acpi_backlight_cmdline = acpi_backlight_native;
> 	if (!strcmp("none", acpi_video_backlight_string))
> 		acpi_backlight_cmdline = acpi_backlight_none;
> }

Yes you are right that should be "acpi_backlight=none".

> and Documentation/admin-guide/kernel-parameters.txt could stand to be updated
> with a few of those options.

Ack, I've put this on my (much too long) TODO list. Feel free to
beat me to it.

Regards,

Hans




> 
> 
>> interface, this will stop e.g. gnome from showing a non working brightness
>> slider in its top right system menu.
>>
>> If this works you can make this permanent and avoid other users from having to
>> do the same thing by writing a patch adding a quirk for this like this one:
>>
>> https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=d21a91629f4b8e794fc4c0e0c17c85cedf1d806c
> 
> 

Re: ACPI Video Driver creates backlight on desktop board

[Randy Dunlap]

On 3/30/20 2:32 PM, Hans de Goede wrote:
> Hi,
> 
> On 3/30/20 11:25 PM, Randy Dunlap wrote:
>> On 3/30/20 1:51 PM, Hans de Goede wrote:
>>> Hi,
>>>
>>> On 3/29/20 11:36 AM, Paul Menzel wrote:
>>>> Dear Linux folks,
>>>>
>>>>
>>>> On the MSI desktop board [1]
>>>>
>>>>       [    0.000000] DMI: Micro-Star International Co., Ltd. MS-7A37/B350M MORTAR (MS-7A37), BIOS 1.MR 12/02/2019
>>>>
>>>> with an AMD Ryzen 3 2200G with Radeon Vega Graphics, the ACPI Video Driver `video` is loaded and creates a backlight device.
>>>>
>>>>       $ readlink -f /sys/class/backlight/acpi_video0
>>>>       /sys/devices/pci0000:00/0000:00:08.1/0000:26:00.0/backlight/acpi_video0
>>>>
>>>> I wonder what the driver is used for as the AMDGPU driver exists for the graphics device.
>>>
>>> Backlight on x86 hw is a bit of a mess, there is an ACPI standard for accessing it
>>> and with older (Windows XP era) laptops that is the interface which usually works,
>>> then there are a bunch of vendor specific SMBIOS or WMI backlight interfaces and
>>> then there is the option of directly accessing the hardware as the amdgpu driver
>>> is doing.
>>>
>>> We have a bunch of heuristics to avoid the acpi_video driver registering a
>>> backlight interface when it should not, either because the direct hw access
>>> should be used instead; or because there simply is no builtin LCD panel and thus
>>> no backlight to control.
>>>
>>> These heuristics are failing on your board.
>>>
>>>> If it’s useful, is there a way to prevent the backlight interface from getting created?
>>>
>>> You can pass "acpi_backlight=ignore" on the kernel commandline to disable the
>>
>> Hi Hans,
>> Should that be "acpi_backlight=none"?  I don't see 'ignore' allowed here:
>>
>> static void acpi_video_parse_cmdline(void)
>> {
>>     if (!strcmp("vendor", acpi_video_backlight_string))
>>         acpi_backlight_cmdline = acpi_backlight_vendor;
>>     if (!strcmp("video", acpi_video_backlight_string))
>>         acpi_backlight_cmdline = acpi_backlight_video;
>>     if (!strcmp("native", acpi_video_backlight_string))
>>         acpi_backlight_cmdline = acpi_backlight_native;
>>     if (!strcmp("none", acpi_video_backlight_string))
>>         acpi_backlight_cmdline = acpi_backlight_none;
>> }
> 
> Yes you are right that should be "acpi_backlight=none".
> 
>> and Documentation/admin-guide/kernel-parameters.txt could stand to be updated
>> with a few of those options.
> 
> Ack, I've put this on my (much too long) TODO list. Feel free to
> beat me to it.

Sure, I'll send a patch shortly.

-- 
~Randy