On Mon, 2004-09-20 at 23:29, Michael Schmitz wrote: > A ghost image of whatever is on the screen, displaced by (roughly) 128 > pixels to the left. At the refresh rate used by X, the ghost image > flickers so that you see it light up with a periodicity of around a cm > vertical, or approximately 40 pixels (that's rather harder to estimate). > The individual streaks of ghosting seem to wander down, then jump up > again. > > Funny enough, it's systematic enough to persist across a suspend to disk > :-)
Ok, this isn't flicker, more like the PLL locking improperly.. I've seen this happening randomly on anton's new laptop, usually it goes away if you force radeonfb to reprogram the display a couple of times. Have you tried, when it happens, turning the backlight all the way down to 0, waiting a bit, then back up ? Does this help at all ? > If you have any ideas about which registers to look at, I'd be happy to > debug this further. With radeonfb debug messages enabled I've only seen > LVDS_GEN_CNTL show any variation between initial boot setup and later > console switches. But that doesn't show a difference between the X vt and > the console vt ... > > > > Maybe time to be more specific WRT sensor and fan location in the driver. > > > > I accept patches :) Would be worth looking more precisely at what Darwin > > does too > > Yep. At one time in the distant past I had registered for access to Darwin > source but that's lost in the noise... I'll re-pull and send a patch, > including the refrigerator support for the thermostat (for pmdisk). > > Michael -- Benjamin Herrenschmidt <[EMAIL PROTECTED]>