Bug 19393 - Crash with two monitors connected
Summary: Crash with two monitors connected
Status: RESOLVED NOTOURBUG
Alias: None
Product: xorg
Classification: Unclassified
Component: Driver/Radeon (show other bugs)
Version: git
Hardware: x86-64 (AMD64) Linux (All)
: medium normal
Assignee: xf86-video-ati maintainers
QA Contact: Xorg Project Team
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2009-01-04 18:09 UTC by Pyotr Berezhkov
Modified: 2009-01-08 02:17 UTC (History)
1 user (show)

See Also:
i915 platform:
i915 features:


Attachments
bug19393-Xorg.0.log (57.90 KB, patch)
2009-01-05 15:56 UTC, Pyotr Berezhkov
no flags Details | Splinter Review
bug19393-xorg.conf (1.79 KB, application/octet-stream)
2009-01-05 15:57 UTC, Pyotr Berezhkov
no flags Details
bug19393-xrandr (Output of xrandr command - note incorrect [1152x768] mode for 1024x768 VGA monitor!) (1.01 KB, text/plain)
2009-01-05 15:59 UTC, Pyotr Berezhkov
no flags Details
bug19393-console-output (4.40 KB, text/plain)
2009-01-05 16:01 UTC, Pyotr Berezhkov
no flags Details

Description Pyotr Berezhkov 2009-01-04 18:09:32 UTC
Display works, and XRandR mode change works, with _one_ monitor (either
VGA or HDMI/DVI) physically connected.  With _both_ monitors connected I
get a black screen and sometimes system crash.

If second monitor (either) is connected _after_ X is running, then display
works, and resizing too.

Hardware:
ASUS M3A78 Pro Motherboard
Athlon X2 5200+ 2700 MHz (Brisbane)
Host bridge: Advanced Micro Devices [AMD] RS780
VGA controller: Radeon HD 3200

Driver:
xf86-video-ati git branch of Tue Dec 30 19:05:58 2008 -0500
(commit c39ca66a)
Comment 1 Alex Deucher 2009-01-04 18:15:07 UTC
Please attach your xorg log and config with both monitors attached.
Comment 2 paul 2009-01-05 06:13:35 UTC
I don't know if this is related, but to a non programmer, it sure seems to be the same pattern as bug 16781, in which X fails when 2 outputs are running.  The difference there is that one of the outputs happens to be the on board LVDS.  According to 16781, the problem seems to be faulty code in atombios causing an endless loop.
Comment 3 Pyotr Berezhkov 2009-01-05 15:56:39 UTC
Created attachment 21702 [details] [review]
bug19393-Xorg.0.log
Comment 4 Pyotr Berezhkov 2009-01-05 15:57:28 UTC
Created attachment 21703 [details]
bug19393-xorg.conf
Comment 5 Pyotr Berezhkov 2009-01-05 15:59:38 UTC
Created attachment 21704 [details]
bug19393-xrandr (Output of xrandr command - note incorrect [1152x768] mode for 1024x768 VGA monitor!)
Comment 6 Pyotr Berezhkov 2009-01-05 16:01:01 UTC
Created attachment 21705 [details]
bug19393-console-output
Comment 7 Pyotr Berezhkov 2009-01-05 16:26:31 UTC
Some corrections and additions to original bug report:

  * VGA monitor displays OK, though it starts up with incorrect mode of
    1152x768 (54.8 Hz), which the monitor (1024x768 pixel) doesn't even
    support!  This can be corrected by invoking:
      'xrandr --output VGA-0 --mode 1024x768'
    Maybe this curious bug provides a key to what's going on?
  * DVI monitor simply goes to sleep and refuses to wake up, but the
    system doesn't crash, as I incorrectly reported in the initial bug
    report.  No 'xset' or 'xrandr' command is able to wake it up.  I
    tried all the possible variations.
  * DVI monitor is connected to HDMI output on motherboard via an
    HDMI/DVI connector (though this hasn't caused problems otherwise).
  * The radeonhd driver doesn't exhibit this bug (though there are other
    issues with it, namely crash on suspend/switch-to-console with the
    VGA output turned _off_).
Comment 8 Pyotr Berezhkov 2009-01-06 10:09:58 UTC
Workaround discovered!

By logging in to my main machine over SSH, I discovered that the
following sequence of commands gets both monitors to display:

  xrandr --output VGA-0 --off
  xrandr --output HDMI-0 --off
  xrandr --output HDMI-0 --auto
  xrandr --output VGA-0 --auto
Comment 9 Alex Deucher 2009-01-07 08:56:41 UTC
The xserver is picking the 1152x768 mode from the default mode pool since it's attempting to match the modes of the two monitors as closely as it can.  This should be handled better on newer versions of the xserver.
Comment 10 Alex Deucher 2009-01-07 08:57:12 UTC
can you test the atom-rework branch of my personal repo?
http://cgit.freedesktop.org/~agd5f/xf86-video-ati/?h=atom-rework
Comment 11 Pyotr Berezhkov 2009-01-08 02:17:37 UTC
Bug closed, as it's not driver-related.


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.