Bug 36062

Created attachment 45397 [details]
Xorg log

Created attachment 45398 [details]
dmesg

Created attachment 45399 [details]
complete xrandr --verbose

Created attachment 45400 [details]
vbios dump

Can you please grab intel_reg_dumper from http://cgit.freedesktop.org/xorg/app/intel-gpu-tools and attach the output?

Created attachment 45408 [details]
output of ./intel_gpu_dump

I added the output of gpu_dump... I guess that's what you meant ?

The only tools I got from gpu-tools were :

intel_gpu_dump intel_gpu_top intel_reg_write intel_stepping

Created attachment 45409 [details]
./intel_reg_dumper output

I got it compiling the git tree...
Nevermind the last comment

Looking at that dump we have a VGA monitor with SSC enabled on the source. Everything else appears relatively sane.

You should be able to test the theory that it is the use of SSC that is causing this by passing i915.lvds_use_ssc=0 on the grub kernel command line.

Created attachment 49089 [details]
EDID Override INF file that corrects Samsung 244T 1920x1200 resolution under Windows

I have the exact same problem in our office using Fedora 15 x64 with a new Dell Inspiron N5110 (Core i7 with Nvidia Optimus dGPU) notebook and Samsung SyncMaster 244T monitor, connected via analog VGA cable (I cannot test the DVI input on the monitor because the notebook does not have DVI out).

I also saw this problem when a colleague connected his Samsung Q330-JS02 (Core i3 with Nvidia Optimus dGPU, Windows 7) notebook to the 244T monitor (again with a VGA cable). I first thought the Optimus feature is causing this, but after some research, I learned that the SyncMaster 244T model has a weirdness in its EDID info that somehow confuses the Intel GPU. Following the HOWTO at this address,

http://www.sevenforums.com/tutorials/7947-force-dvi-hdmi-resolutions-refresh-rates.html

I was able to create the attached INF file with a custom EDID for Windows, and once loaded it fixed the issue my colleague's notebook.

I've been trying to do something similar in Fedora for the Dell N5110, but with KMS and all the display auto-detection going on, I couldn't figure out how I can override the EDID for the Intel driver. I've read that the Nvidia driver has an option for loading a custom EDID, but thanks to Optimus technology, I cannot use the Nvidia GPU under Linux.

Does the Intel driver provide a similar option for EDID override? If possible, feeding the modified EDID in the attached INF file to the Intel driver might fix this issue.

commit 199e5d79f1c988a8039fa75b736a3adcdda56abc
Author: Keith Packard <keithp@keithp.com>
Date:   Thu Sep 22 12:01:57 2011 -0700

    drm/i915: Fix PCH SSC reference clock settings
    
    The PCH refclk settings are global, so we need to look at all of the
    encoders, not just the current encoder when deciding how to configure
    it. Also, handle systems with more than one panel (any combination of
    PCH/non-PCH eDP and LVDS).
    
    Disable SSC clocks when no panels are connected.
    
    Signed-off-by: Keith Packard <keithp@keithp.com>

Hi,
   I just tested with i915 Version: 2:2.20.9-0ubuntu2  srcversion :1261F63CFAFFBE7E5D8B95B on the latest 12.10 ubuntu.

  And the problem is still present, I also tested disabling ssc but had the same problem...

  Ideas ?

Thanks

We don't have a way to load a whole EDID into the kernel (though X used to have an option for that), but you can create custom modes based on the INF file and use them.

(In reply to comment #14)
> We don't have a way to load a whole EDID into the kernel (though X used to
> have an option for that), but you can create custom modes based on the INF
> file and use them.

We've _had_ no way, now fixed with the EDID firmware interface ...

Author: Carsten Emde <C.Emde@osadl.org>
Date:   Sun Mar 18 22:37:33 2012 +0100

    drm: allow loading an EDID as firmware to override broken monitor

On Tue, 11 Dec 2012 21:46:36 +0000
bugzilla-daemon@freedesktop.org wrote:

> https://bugs.freedesktop.org/show_bug.cgi?id=36062
> 
> --- Comment #15 from Daniel Vetter <daniel@ffwll.ch> ---
> (In reply to comment #14)
> > We don't have a way to load a whole EDID into the kernel (though X used to
> > have an option for that), but you can create custom modes based on the INF
> > file and use them.
> 
> We've _had_ no way, now fixed with the EDID firmware interface ...
> 
> Author: Carsten Emde <C.Emde@osadl.org>
> Date:   Sun Mar 18 22:37:33 2012 +0100
> 
>     drm: allow loading an EDID as firmware to override broken monitor
> 

Ooh fancy...

Hardware bug, to override the EDID requires loading it through the kernel drm_kms_helper module parameters.