Monitor in need of EDID quirk
Submitted by Stuart Bennett
Assigned to Xorg Project Team
Description
Created attachment 19379 Raw EDID (gzip'd, as firefox crashes when I attempt to attach it otherwise...)
The attached EDID is a bit broken, reporting a 1280x960@100Hz resolution in the Standard Timings section; offset 0x26 is 0x81, 0x68:
0x81 * 8 + 248 == 1280 px width (0x68 & 0xc0) >> 6 == 1 => 4:3 aspect ratio 0x68 & 0x3f == 0x28 == 40, 40 + 60 == 100Hz
Unfortunately the monitor is not even slightly capable of this (max of 1280x1024@60), nor any of the modes which Xorg extrapolates from the 1280x960@100Hz mode.
As this monitor is EDID rev 1.0 it was hypothesised on #xorg-devel that the meaning of bit 5 in the second byte differed under 1.0 parsing, leading to a credible 1280x960@68Hz mode. Sadly, the attached scan from the 1.0 spec shows this not to be a legal interpretation.
The nVidia binary driver appears to work around the problem, by parsing it as a
1280x960@68Hz mode (seen when using -logverbose <lots>
). I'll attach two possible patches which quirk it for this monitor, both against server-1.5-branch (but should apply easily enough to master); one which seems to fit better with the general quirking mechanism, and another which corrects it earlier so xf86PrintEDID
doesn't print misleading pre-quirking info.
Attachment 19379, "Raw EDID (gzip'd, as firefox crashes when I attempt to attach it otherwise...)":
rawedid.gz