On Sun, 26 Aug 2012 15:45:42 +0200, Alon Levy <[email protected]> wrote: > On Sun, Aug 26, 2012 at 12:51:44PM +0100, Chris Wilson wrote: > > On Sun, 26 Aug 2012 21:46:01 +1000, Dave Airlie <[email protected]> wrote: > > > On Sun, Aug 26, 2012 at 9:43 PM, Alon Levy <[email protected]> wrote: > > > > ACPI is meant as a fallback, so it should be last. > > > > > > eh no, acpi is never the fallback, its always the preferred way. > > I guess I was confused by the comment: > "acpi_video1", /* finally fallback to the generic acpi drivers */ > > > > > Right, acpi works on far more machines than intel_backlight. If you want > > to override the automaticlly choose interface, how about an > > Option "backlight" "intel_backlight" > > ? > > Works for me, patch below. But I'm confused - wouldn't the > intel_backlight sys directory be non existant if the kernel found the > machine didn't support it? at least I thought that was the point - then > checking acpi last makes sense.
It's worse than you thought. We know the hardware always has PWM registers that may be used by the backlight controller. However, the manufacturer is at liberty to implement their own controller entirely and expose that interface through ACPI. So on quite a few machines intel_backlight does nothing at all, and on other machines the ACPI firmware is worse - but the kernel has no way of telling just which path is broken. Thanks for the patch, -Chris -- Chris Wilson, Intel Open Source Technology Centre _______________________________________________ [email protected]: X.Org development Archives: http://lists.x.org/archives/xorg-devel Info: http://lists.x.org/mailman/listinfo/xorg-devel
