Jesse Barnes jbarnes at virtuousgeek.org
Thu Nov 29 11:14:26 PST 2007

On Thursday, November 29, 2007 11:10 am Colin Guthrie wrote:
> Jesse Barnes wrote:
> > On Wednesday, November 28, 2007 1:08 am Lukas Hejtmanek wrote:
> >> On Sat, Nov 24, 2007 at 03:34:57AM +0000, Sergio Monteiro Basto wrote:
> >>> Lukas, I see 3 patches in this tread, what is the best combination ?
> >>> "also something" say that is this patch more something, should I apply
> >>> all of the 3 or just yours 2 ?
> >>> I have applied yours 2 and no crash nor flicks.
> >>
> >> You should apply all 3 patches to have the backlight behaving mostly
> >> correct.
> >
> > Does this patch work for you guys?
> >
> > It properly gets the backlight at startup time, saves and restores it
> > across DPMS on/off calls, and doesn't mess with it at startup time (iow
> > it assumes you have it where you want it when you start X).
> If I set the backlight control to native via xrandr (as you pointed out
> in another thread). Then my backlight works fine with these patches, but
> then I'm not certain how badly it was broken before as I didn't try
> native mode!
> Under kernel mode (the default) the backlight doesn't appear to function
> too well. Previously it was maxed out, but after the patches it was
> minimal but to be honest I'm not sure that's 100% correct as it was all
> quite random and it flipped round when power was removed etc.

Hm, yeah I've seen some weirdness with the kernel backlight drivers too, but 
it seems to be independent of X, so I'll assume it's not my problem until 
prooven otherwise. :)

> But like I say with native mode and these patches it seems to work fine,
> although one thing that's interesting is that the BACKLIGHT level
> doesn't change in xrandr --prop regardless of the actual light level...
> it always says:
>         BACKLIGHT: 624 (0x00000270) range:  (0,624)

Yeah, that's a server limitation, we need to add an actual callback into the 
driver when the user asks for the current value.  Right now it just returns 
the last requested value.


More information about the xorg mailing list