On Thu, 14 Mar 2013, Chris Li <lkml at chrisli.org> wrote: > Hi, I attach the two demsg and with and without the bad commit > on the intel nightly branch. Without the bad commit it actually works. > However, on the tip of intel nightly. the moeset work around does not work > there any more. Hi Chris, thanks for the dmesgs. So the diff in the dmesgs from intel nightly (bad) to the same with a57f7f9175b8ccbc9df83ac13860488913115de4 reverted (good) wrt backlight boils down to this, repeated a few times: -[drm:asle_set_backlight], bclp = 0x80000000 +[drm:asle_set_backlight], bclp = 0x800000ff [drm:intel_panel_get_max_backlight], max backlight PWM = 4648 -[drm:intel_panel_actually_set_backlight], set backlight PWM = 0 -[drm:asle_set_backlight], bclp = 0x80000000 +[drm:intel_panel_actually_set_backlight], set backlight PWM = 4648 +[drm:asle_set_backlight], bclp = 0x800000ff [drm:intel_panel_get_max_backlight], max backlight PWM = 4648 -[drm:intel_panel_actually_set_backlight], set backlight PWM = 0 +[drm:intel_panel_actually_set_backlight], set backlight PWM = 4648 Fun. The BIOS seems to ask for zero backlight. Maybe it means something else for Windows 8. White is the new black or something. > Let me know if you need any thing else from me. Just one idea, please attach hexdumps of /sys/kernel/debug/dri/0/i915_opregion (with debugfs mounted there, obviously) for both kernels. Maybe we're missing something there that the BIOS thinks Windows 8 should handle. I've never used the acpi_osi= kernel parameter, but it looks like you could workaround this with acpi_osi="!Windows 2012". Please check that running the "bad" kernel. BR, Jani.
