(some of you might get this mail in double copy , sorry)

On 2/11/07, Luca Tettamanti [off-list ref] wrote:

Sorry for the delay.

Ditto!

It also seemed that my kernel compiling sk1llz had gone AWL, I
couldn't get the newly compiled kernel to run, until I realized the
initrd.img was mislinked. Anyway.

Your patch has an immediately visible effect of reducing snow: it
seems that now that you fixed the access bug the EDID is properly
/not/ used. dmesg | grep nvidiafb now claims the following:

"""

nvidiafb: Device ID: 10de0112
nvidiafb 0000:01:00.0: Unable to read EDID block.
nvidiafb 0000:01:00.0: Unable to read EDID block.
nvidiafb 0000:01:00.0: Unable to read EDID block.

(it actually gave the "I2C probe failed ..." message on console, but
it's not appearing in the dmesg, obviously, since it's a simple
printk)

nvidiafb 0000:01:00.0: Unable to read EDID block.
nvidiafb 0000:01:00.0: Unable to read EDID block.
nvidiafb 0000:01:00.0: Unable to read EDID block.
nvidiafb: CRTC 1 is currently programmed for DFP
nvidiafb: Using DFP on CRTC 1
nvidiafb: Panel size is 1600 x 1200
nvidiafb: Panel is TMDS
nvidiafb: MTRR set to ON
nvidiafb: Flat panel dithering disabled
nvidiafb: PCI nVidia NV11 framebuffer (32MB @ 0xE0000000)

"""

So the EDID still seems to be totally inaccessible (which means I
can't really tell why/when/where it's b0rked).

Finally, as I mentioned, console still has a little snow. It's barely
perceptible in common console usage because the monitor is mostly
black, but can get annoying with fullscreen apps à la mc.

This extra snow *might* be the effect of bandwidth limits Antonino
Daplas was referring to in this lkml message:
http://lkml.org/lkml/2005/10/1/7 but I'll have to enquire on this
further.

-- 
Giuseppe "Oblomov" Bilotta