[RS690] (Needs new -ati) Wrong resolution and Nokia 800 Pro+

This seems to be a bug in the xserver-xorg-video-radeon driver becasue with xserver-xorg-video-radeonhd I get the full resolution (though other bugs like a wrong frequency of only 60 MHz instead of the needed 108 MHz. I'll open another bug for that issue.

About the monitor: I was able to get this page from the Google cache, seems like Viewsonic (who acquired the Nokia monitor line) removed the page
http://www.viewsonic.com/support/desktopdisplays/lcddisplays/nokia/800pro/

Interesting bits from that page:

Features
 * Horizontal Scan Frequency: 30-82 kHz
 * 1280x1024 optimum resolution
 * Pixel Pitch: 0.28mm
 * Color: 16.7 Million True Color
 * Viewing Angle: 160° Horizontal and Vertical
 * Multimedia: BusinessAudio System
 * Input Capability: 30 Pin P & D Connection and 15 Pin Mini D-Sub
 * USB Connectivity: 1 Upstream Port, 4 Downstream Ports
 * Plug & Play +
 * ISO 9001 and ISO 14001

Specifications
  LCD Panel
  * Type 18.1" VIS
  * True Resolution 1280x1024
  * Contrast Ratio 300:1
  * Viewing Angle 160° horizontal, 160° vertical
  * Brightness 235 cd/m2 typical
  * Panel Surface SafetyPlex screen
  INPUT SIGNAL
  * Sync H/V sync (TTL)
  * Analog RGB Analog (0.7 Vpp, 75 ohms)
  * Frequency Fh:29-82kHz, Fv:55-86Hz
  * Bandwidth 140MHz
  CONNECTOR
  * Analog 15-pin mini D-sub
  * Power External AC/DC single phase, 6 ft detachable power cord
  POWER
  * Voltage AC 100-240 VAC 47-63Hz
  * Consumption Three-year limited warranty on parts, labor and back light.

And for completeness, here's the xrandr output and the Xorg.0.log:

mss@Otherland:~$ xrandr
Screen 0: minimum 320 x 200, current 1280 x 1024, maximum 1440 x 1440
VGA_1 disconnected
DVI-D_1 disconnected
DVI-D_2 connected 1280x1024+0+0 359mm x 287mm
   1280x1024 60.0*+
   1440x900 59.9
   1280x960 60.0
   1360x768 59.8
   1152x864 60.0
   1024x768 60.0
   800x600 60.3
   640x480 60.0 59.9

fglrx gets it right:

mss@Otherland:~$ xrandr
Screen 0: minimum 320 x 200, current 1280 x 1024, maximum 1280 x 1024
default connected 1280x1024+0+0 0mm x 0mm
   1280x1024 60.0*
   1280x960 60.0
   1280x768 60.0
   1280x720 60.0
   1152x864 60.0
   1024x768 60.0
   800x600 60.0
   720x480 60.0
   640x480 60.0
   640x400 60.0
   512x384 60.0
   400x300 60.0
   320x240 60.0
   320x200 60.0

Hmm, you're right that it's interesting that the resolution is being detected differently by each of the three drivers, and the inconsistency may indeed point to some sort of issue with -ati; I'll upstream it to them for that reason.

However, from your description it seems that none of the drivers get it quite perfect. I note that both -ati and -fglrx are seeing the physical screen size as 0x0 mm, which suggests bad EDID; yet -radeonhd is getting valid looking measurements. But you say that radeonhd is getting the wrong refresh rate, which suggests that even it isn't quite perfect.

My guess is that your monitor provides EDID that is just slightly out of wack from what the xserver expects, and it's mis-parsing it. Perhaps each of the drivers has some correction code that handles the error condition in a slightly different way, thus causing the range of different results.

Notice how the h/v rates don't match the vendor's advertised capabilities:

Via EDID:
 HorizSync 31-65
 VertRefresh 58-62

Via DDC:
 monitorrange: 31-65, 58-62

Xorg.0.log (radeonhd):
  (II) RADEONHD(0): Ranges: V min: 58 V max: 62 Hz, H min: 31 H max: 65 kHz, PixClock max 110 MHz

Advertised specifications:
  * Frequency Fh:29-82kHz, Fv:55-86Hz

So, what happens if you plug in these values into your xorg.conf?

Section "Monitor"
    ...
    HorizSync 29 - 82
    VertRefresh 55 - 86
EndSection

We're closing this bug since it is has been some time with no response from the original reporter. However, if the issue still exists please feel free to reopen with the requested information. Also, if you could, please test against the latest development version of Ubuntu, since this confirms the bug is one we may be able to pass upstream for help.

Sorry, I didn't have time to investigate further (except that adding the lines you wrote didn't help). fglrx worked fine so I didn't have to work on this issue.

Until today I upgraded to jaunty where fglrx is obiously broken. Bad news: After I removed every trace of fglrx from my system, both radeon and radeonhd are broken as well. radeonhd displayed a distorted screen, radeon nothing at all.

I'll attach the Xorg.logs, both with and without the config lines you posted.

I must correct myself: radeonhd works fine, only the radeon driver fails.

I originally first removed the fglrx driver, rebooted, radeonhd was autoselected (and failed), uninstalled radeonhd and while doing so noticed that I forgot to remove some fglrx-mesa-stuff. After removing both packages I tried radeon and it failed as described above. Now I re-added radeonhd and everything seems to be fine, looks almost better than the fglrx driver at the first glance.

Section "Screen"
 Identifier "Screen"
 Device "Device"
EndSection

Section "Device"
 Identifier "Device"
 Driver "radeon"
 Option "IgnoreEDID" "true"
 Option "NoDDC" "true"
EndSection

Section "Monitor"
 Identifier "HDMI-0"
 Option "Enable" "false"
# Option "Ignore" "true"
EndSection

Section "Monitor"
 Option "Enable" "true"
 Option "PreferredMode" "1280x1024"
        # Block type: 2:0 3:fd
        # Block type: 2:0 3:ff
        # Block type: 2:0 3:fc
        Identifier "DVI-0"
        VendorName "NOK"
        ModelName "NOKIA 800PRO"
        # Block type: 2:0 3:fd
        HorizSync 31-65
        VertRefresh 58-62
        # Max dot clock (video bandwidth) 110 MHz
        # Block type: 2:0 3:ff
        # Block type: 2:0 3:fc
        # DPMS capabilities: Active off:yes Suspend:yes Standby:yes

        Mode "1280x1024" # vfreq 60.020Hz, hfreq 63.981kHz
                DotClock 108.000000
                HTimings 1280 1328 1440 1688
                VTimings 1024 1025 1028 1066
                Flags "+HSync" "+VSync"
        EndMode
        # Block type: 2:0 3:fd
        # Block type: 2:0 3:ff
        # Block type: 2:0 3:fc
EndSection

Fixed upstream in http://cgit.freedesktop.org/xorg/driver/xf86-video-ati/commit/?id=e01b8dc827ce1913927e1a4fad05a99a2dd9e805

Excellent. Thanks for reporting and testing upstream.

Patch does not directly apply to 6.12.2 codebase. We'll pull the fix in when we update -ati to latest in karmic.

The patch is actually quite small, I filtered the two needed hunks from the big diff. Here's a debdiff if you're interested :) So far this patched driver seems to work fine.

Excellent, that does the trick. Thanks, upload sponsored.

This bug was fixed in the package xserver-xorg-video-ati - 1:6.12.2-0ubuntu3

---------------
xserver-xorg-video-ati (1:6.12.2-0ubuntu3) karmic; urgency=low

  * Add 106_fix_dvi_on_rs690.patch: Fix DCI output on some rs690
    cards. Filtered from upstream.
    (LP: #298094)

 -- <email address hidden> (Malte S. Stretz) Tue, 05 May 2009 17:25:01 +0200