https://bugzilla.kernel.org/show_bug.cgi?id=73521

Aaron Lu aaron.lu@intel.com changed:

What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |ASSIGNED CC| |aaron.lu@intel.com Component|Other |Video(DRI - non Intel) Assignee|acpi_other@kernel-bugs.osdl |drivers_video-dri@kernel-bu |.org |gs.osdl.org Product|ACPI |Drivers

--- Comment #1 from Aaron Lu aaron.lu@intel.com --- (In reply to patrick.clara from comment #0)

Created attachment 131421 [details] dsdt.dls

Hello.

On the asus ux51vz laptop screen backlight is not working. It is a win8 laptop and has only one single nvidia gpu.

By default in /sys/class/backlight i have acpi_video0. Pressing the hotkeys actually result in a change in acpi_video0/actual_brightness. Also writing manually to acpi_video0/brightness results in a change in acpi_video0/actual_brightness. Apparently everything works except that the actual brightness of the screen doesnt change. Looking at the the _BCM method in the DSDT i saw that it calls STBR which does nothing it think. So the ACPI is broken in my case.

agree.

I also tested video.use_native_backlight=1 but acpi_video0 still got registred, maybe becouse of the backlight_device_registered(BACKLIGHT_RAW) in this check:

if (acpi_osi_is_win8() && use_native_backlight && backlight_device_registered(BACKLIGHT_RAW))

right.

If i understand correctly, since the only raw backlight device going to be registed is nv_backlight, and since nouveau requires the video module and therefore gets loaded afterwards, this condition will never become true.

I tried removing the backlight_device_registered(BACKLIGHT_RAW) and than video.use_native_backlight=1 do what it is supposed to do.

The problem remains in the way asus-wmi and nouveau check if they should register their own control devices. They use acpi_video_backlight_support() I think which is unaffected by the use_native_backlight parameter.

We will need to talk to nv gpu developers on this. I don't think the following is the correct behaviour - whether ACPI video has provided an interface doesn't matter if the gpu driver should provide a control interface, the only criteria should be: the interface is not broken.

#ifdef CONFIG_ACPI if (acpi_video_backlight_support()) { NV_INFO(drm, "ACPI backlight interface available, " "not registering our own\n"); return 0; } #endif

Just for testing I tried removing the check from the nouveau driver and forcing it to register the nv_backlight. This way, booting with use_native_backlight=1 actually gives a system with working backlight, altough this solution is far from beeing optimal and also far from beeing out of the box.

We can easily add APIs in acpi video module to unregister its backlight interface without losing its event reporting capability. But we will need to talk to nv gpu developers on this first - is it the case for win8 systems, the gpu is in control of the backlight? For many intel gpu based system it is the case.