Bug 15175 - radeon rs690 issues with 1920x1200 DVI (blanking)
Summary: radeon rs690 issues with 1920x1200 DVI (blanking)
Status: RESOLVED DUPLICATE of bug 22229
Alias: None
Product: xorg
Classification: Unclassified
Component: Driver/Radeon (show other bugs)
Version: git
Hardware: Other All
: medium normal
Assignee: xf86-video-ati maintainers
QA Contact: Xorg Project Team
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2008-03-23 01:37 UTC by Marko Macek
Modified: 2009-06-11 09:49 UTC (History)
3 users (show)

See Also:
i915 platform:
i915 features:


Attachments
xorg log (125.83 KB, patch)
2008-03-23 01:37 UTC, Marko Macek
no flags Details | Splinter Review
xorg conf (1.16 KB, application/octet-stream)
2008-03-23 01:38 UTC, Marko Macek
no flags Details
xorg log 2 (82.38 KB, text/plain)
2008-03-23 11:46 UTC, Marko Macek
no flags Details
xorg conf 2 (1012 bytes, text/plain)
2008-03-23 11:48 UTC, Marko Macek
no flags Details
possible fix for 1920x1200 mode (443 bytes, patch)
2008-03-24 14:18 UTC, Alex Deucher
no flags Details | Splinter Review
xv clipping problem (tvtime) (634.02 KB, image/png)
2008-03-27 11:03 UTC, Marko Macek
no flags Details
possible fix (937 bytes, patch)
2009-02-26 12:07 UTC, Alex Deucher
no flags Details | Splinter Review

Description Marko Macek 2008-03-23 01:37:14 UTC
Created attachment 15394 [details] [review]
xorg log

from git (6d5066a451017a2683addc9e2496987626795dda)

blog comments:
--
I have RS690 and have 2 problems

- I have resolution of 1920×1200 (DVI) and my screen goes blank randomly for several seconds. Is there a solution for this (If I use rX550, no problems -> switching back)

- The latest git version seems to be much slower when Composite is disabled (I haven’t tried enabling it yet) than one in FC9 rawhide

- although Xv works fine unless the window is partially covered (on the left side), clipping seems to subtract from both sides although only one is covered.

--

slowness is probably 'cause DRI doesn't work (gears 400fps)
Comment 1 Marko Macek 2008-03-23 01:38:35 UTC
Created attachment 15395 [details]
xorg conf
Comment 2 Marko Macek 2008-03-23 06:12:26 UTC
could the bug 14810 be related to issue #1?
Comment 3 Alex Deucher 2008-03-23 07:57:03 UTC
(In reply to comment #2)
> could the bug 14810 be related to issue #1?
> 

What distro are you using?  That bug seems to be related to some desktop service polling the outputs with randr every few seconds.  Do you still get the blanking problems if you start just plain plain X server with no desktop environment (run Xorg rather than startx).
Comment 4 Marko Macek 2008-03-23 11:32:00 UTC
Distro is current rawhide (upgraded from fc8 + current git)

I still get random blanking, after killing the usual suspects (dbus, udev, ...)

gears is running fullscreen at 17fps, I'm not sure about DRI working, it says:

Warning, RS690 detected, 3D support is incomplete.


This is my process list, except kernel stuff:


root      2333     1  0 18:18 ?        00:00:00 /usr/sbin/sshd
xfs       2450     1  0 18:18 ?        00:00:00 xfs -droppriv -daemon
root      2577     1  0 18:18 tty4     00:00:00 /sbin/mingetty tty4
root      2578     1  0 18:18 tty5     00:00:00 /sbin/mingetty tty5
root      2579     1  0 18:18 ?        00:00:00 login -- mark     
root      2580     1  0 18:18 tty3     00:00:00 /sbin/mingetty tty3
root      2581     1  0 18:18 ?        00:00:00 login -- root     
root      2583     1  0 18:18 tty6     00:00:00 /sbin/mingetty tty6
root      2813  2581  0 18:19 tty1     00:00:00 -bash
mark      2857  2579  0 18:20 tty2     00:00:00 -bash
mark      2927  2857  0 18:20 tty2     00:00:00 /bin/sh /usr/bin/startx
mark      2945  2927  0 18:20 tty2     00:00:00 xinit /etc/X11/xinit/xinitrc -- -auth /home/mark/.serverauth.2927
root      2946  2945 32 18:20 tty7     00:22:06 X :0 -auth /home/mark/.serverauth.2927
mark      2951  2945  0 18:20 ?        00:00:00 icewm-session
mark      3110  2951  0 18:20 ?        00:00:00 /usr/bin/ssh-agent /home/mark/.Xclients
mark      3159  2951  0 18:20 ?        00:00:04 [xterm]
mark      3183  3159  0 18:20 ?        00:00:00 xterm
mark      3185  3183  0 18:20 pts/0    00:00:00 bash
mark      3241  3159  0 18:21 ?        00:00:01 xterm
mark      3243  3241  0 18:21 pts/1    00:00:00 bash
root      3530  3243  0 18:26 pts/1    00:00:00 bash
mark      3571  3185  1 18:27 pts/0    00:00:56 glxgears
mark      3630  3159  0 18:32 ?        00:00:00 /bin/sh /opt/firefox/firefox
mark      3633  3630  0 18:32 ?        00:00:00 /bin/sh /opt/firefox/run-mozilla.sh /opt/firefox/firefox-bin
mark      3639  3633  8 18:32 ?        00:04:33 /opt/firefox/firefox-bin
mark      3713  3159  0 18:36 ?        00:00:00 xterm
mark      3715  3713  0 18:36 pts/2    00:00:00 bash
mark      3747     1  0 18:36 ?        00:00:00 /usr/libexec/gconfd-2 12
mark      3763  3159  0 18:41 pts/3    00:00:00 bash
mark      3815  3763  0 19:01 pts/3    00:00:01 icewm
mark      3816  3815 11 19:05 ?        00:02:28 tvtime
mark      3817  3815  0 19:12 ?        00:00:00 rxvt
mark      3818  3817  0 19:12 pts/4    00:00:00 bash
root      3857  3530  5 19:27 pts/1    00:00:00 ps -ef

If the system is idle (except for gears/tvtime) , there is almost no blanking, but as soon as I start doing anything, blanking occurs often.


Perhaps unrelated: 

I got this error once in glxgears (I also got this on X550 once, but it locked up immediately there)

*********************************WARN_ONCE*********************************
File r300_mem.c function r300_mem_alloc line 225
Ran out of GART memory (for 1048576)!
Please consider adjusting GARTSize option.
***************************************************************************
Comment 5 Marko Macek 2008-03-23 11:46:46 UTC
Created attachment 15399 [details]
xorg log 2
Comment 6 Marko Macek 2008-03-23 11:48:03 UTC
Created attachment 15400 [details]
xorg conf 2

new xorg config (prev had some unwanted modeline stuff)
Comment 7 Alex Deucher 2008-03-24 14:18:50 UTC
Created attachment 15434 [details] [review]
possible fix for 1920x1200 mode

Does this patch help with the 1920x1200 mode?
Comment 8 Marko Macek 2008-03-25 11:32:48 UTC
It does not help :(
Comment 9 Alex Deucher 2008-03-27 07:06:38 UTC
Another user reported a modeline with a lower dot clock worked perfectly on his monitor while the standard one from his monitor did not.

Modeline  "1920x1200-dvi"  150.0   1920 1940 2007 2027   1200 1201
1204 1233   +hsync -vsync
Comment 10 Marko Macek 2008-03-27 10:57:58 UTC
Adding a new modeline and switching to it with xrandr doesn't help.

I have also tried on the same display:
1600x1200 -> blinking / loss of sync
1900x1080 -> no loss of sync

I guess dual link DVI is to be having problems.

Maybe HDMI extender card + HDMI cable is worth trying (need to buy the cable)
Comment 11 Marko Macek 2008-03-27 11:03:20 UTC
Created attachment 15515 [details]
xv clipping problem (tvtime)

attaching an image showing the the Xv clipping problems when 
the window is covered.
Comment 12 Marko Macek 2008-03-27 11:05:58 UTC
re: comment #10

could the problem be that the port is identified as HDMI-0

The board actually has a DVI connector which is connected with a DVI->HDMI cable to the monitor.

Under radeon x550, the connector is identified as DVI-0

xrandr output:

Screen 0: minimum 320 x 200, current 1920 x 1080, maximum 1920 x 1200
VGA-0 disconnected (normal left inverted right x axis y axis)
S-video disconnected (normal left inverted right x axis y axis)
HDMI-1 disconnected (normal left inverted right x axis y axis)
HDMI-0 connected 1920x1080+0+0 (normal left inverted right x axis y axis) 432mm 
x 324mm
   1920x1200      60.0 +
   1920x1200-dvi   60.0  
   1920x1080      59.9* 
   1600x1200      60.0     60.0  
   1680x1050      60.0     60.0     59.9  
   1600x1024      60.2  
   1400x1050      74.8     70.0     70.0     60.0  
   1280x1024      75.0     75.0     60.0     60.0  
   1440x900       59.9  
   1280x960       75.0     60.0  
   1360x768       59.8     60.0  
   1280x800       75.0     70.0     60.0  
   1152x864       75.0     75.0     70.0     60.0  
   1280x768       75.0     70.0     60.0  
   1280x720       75.0     70.0     60.0  
   1024x768       75.1     75.0     70.1     60.0  
   832x624        74.6  
   800x600        72.2     75.0     60.3     56.2  
   640x480        75.0     72.8     75.0     60.0     59.9  
   720x400        70.1  
Comment 13 Alex Deucher 2008-03-27 11:36:14 UTC
(In reply to comment #12)
> re: comment #10
> 
> could the problem be that the port is identified as HDMI-0

Those are just names bases on the bios connector tables.  The driver handles the actual ports properly.
Comment 14 Alex Deucher 2008-12-03 01:24:11 UTC
Is this still an issue with a newer driver (6.9.0 or newer)?
Comment 15 Oliver Münch 2008-12-15 12:29:02 UTC
I have the same problems (screen goes blank randomly for
several seconds) with a Sapphire X1650 Pro on a gentoo system, using the 6.9.0 drivers. The blanking occurs only at the 1920x1200 resolution. Changing with xrandr to 1280x1024 for example, there is no blinking at all.
Comment 16 Marko Macek 2008-12-16 11:21:25 UTC
I haven't tried this in a while (the box is now my headless firewall box).

Some time ago I also tried the radeonhd driver, it had the same problem with 1920x1200 resolution.

Not tested with fglrx or MSWindows.

The monitor is now on my current box, running radeon driver:

ATI Technologies Inc RV370 [Sapphire X550 Silent] (prog-if 00 [VGA controller])

which has no problems with blanking (also running 1920x1200 DVI->HDMI).

I will try to retest this over the weekend using Fedora 10 (if that's not enough, I will try to find the time to build from git)
Comment 17 Marko Macek 2008-12-21 09:21:14 UTC
retested with Fedora 10

- fb (drm) and radeon still have the problem. They also both behave strangely the same:
a) the screen clocks are set for 1920x1200 resolution
b) usable resolution is set to 800x600 (console and X shown in top left of the scren) and has to be manually extended in xrandr
the screen going black occurs both in fb and X.

- radeonhd WORKS (so far at least), but no drm and/or Xv

- radeon (textured video) still has the clipping problem the top part of the window (full width) is covered by another window: the content shifts down, and the bottom is doubly clipped (so the height is zero when window is half covered)
This problem was present in horizontal direction before, but now seems ok.
Comment 18 cooper 2008-12-29 00:56:55 UTC
For HDMI radomly flicker issue, I can duplicate it and I debug it for 2 days, no result. Or more accurately, this issue only occur on DVI->HDMI card, if I attach the monitor to HDMI on board directly, 1920x1200 works well, no flicker.

To be more precise, if monitor is connected to DVI->HDMI card, even 1280x1024, monitor will flicker also on a low repetition rate.

So I think maybe it's related to DVI->HDMI card, anyway, in Windows it works well.
Comment 19 Bryce Harrington 2009-02-20 19:22:52 UTC
I have also been seeing an infrequent blanking on my pair of BENQ LCD 1920x1200 panels, pretty much since I originally got them.  I've seen it on 2 different ATI cards (one a 6xx, another 5xx), from the 6.8 driver (maybe earlier) and still see it on 6.11.0.

I hadn't reported it before because I assumed it was a hardware issue, but after reading several bug reports about this including this one and a couple in Ubuntu's launchpad tracker, I think it may be something with the xserver or driver.

One thing I noticed is it seems to only occur on DVI.  Originally I ran it on a dual DVI/VGA card, and I saw the blanking only on the screen connected to DVI.  I then moved to a different DVI/DVI card, and now see the problem on both monitors, although not at the same time.



Comment 20 Bryce Harrington 2009-02-20 19:25:39 UTC
For reference, here are two ubuntu bug reports that seem to have the same issue:
https://bugs.edge.launchpad.net/ubuntu/+source/xserver-xorg-video-ati/+bug/321137
https://bugs.edge.launchpad.net/ubuntu/+source/xserver-xorg-video-ati/+bug/310864
Comment 21 Alex Deucher 2009-02-26 12:07:34 UTC
Created attachment 23331 [details] [review]
possible fix

the monitor may not like the frequency that's getting picked.  let's try tweaking it a bit.  Can you try the attached patch against ati git master?  There are two options to try with the patch: tweaking the frequency up, and tweaking it down.  Please try both combinations by commenting out the appropriate line from the patch below:
clk = mode->Clock - 10;
clk = mode->Clock + 10;
Comment 22 Alex Deucher 2009-06-11 09:49:32 UTC
This should be fixed in git master as of d095bb3157faaee10b0130ea9b1032e0b533aaee.

*** This bug has been marked as a duplicate of bug 22229 ***


Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.