Bug 25460 - dual head not working on sapphire radeon 4670 (HDMI, DVI)
Summary: dual head not working on sapphire radeon 4670 (HDMI, DVI)
Status: RESOLVED FIXED
Alias: None
Product: xorg
Classification: Unclassified
Component: Driver/Radeon (show other bugs)
Version: 7.4 (2008.09)
Hardware: Other All
: medium normal
Assignee: xf86-video-ati maintainers
QA Contact: Xorg Project Team
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2009-12-05 05:10 UTC by Aljaž Prusnik
Modified: 2009-12-25 09:48 UTC (History)
0 users

See Also:
i915 platform:
i915 features:


Attachments
no additional settings in xorg (just radeon driver under driver section) (69.00 KB, patch)
2009-12-05 05:11 UTC, Aljaž Prusnik
no flags Details | Splinter Review
This is the log when I only attach on a HDMI connector (correct behaviour) (49.35 KB, patch)
2009-12-05 05:12 UTC, Aljaž Prusnik
no flags Details | Splinter Review
this is when all other options in xorg are uncommented (both monitor and screen sections) (69.07 KB, patch)
2009-12-05 05:13 UTC, Aljaž Prusnik
no flags Details | Splinter Review
Xorg configuration file for the default xorg.log (1.67 KB, patch)
2009-12-05 05:15 UTC, Aljaž Prusnik
no flags Details | Splinter Review
dmesg, default xorg .conf (only radeon driver) (54.92 KB, patch)
2009-12-05 11:31 UTC, Aljaž Prusnik
no flags Details | Splinter Review
xorg log radeon.modesetting=0, dualhead enabled (168.13 KB, application/octet-stream)
2009-12-05 15:53 UTC, Aljaž Prusnik
no flags Details

Description Aljaž Prusnik 2009-12-05 05:10:04 UTC
I just bought Sapphire 4670 Ultimate with DVI, HDMI and VGA output and replaced a 4350 dual DVI card.
I tried activate dual head but apparently, from what I gather the driver or xorg does not consider that I have 2 monitors it only ever defaults settings to the monitor that is connected to the DVI and then forces the same settings to HDMI connected monitor.
If I try to via randr or gnome display settings to uncheck mirror (or say right-of under randr) then on both monitors I see the right-of part of the display. It looks as though the left screen (HDMI one) does not get the proper output.

Under windows 7 displays are correctly layed out automatically (HP left - basic, Philips right-of - extended) so the card is able to do that.

Also the previous dual DVI card (4350) could properly handle dual head with these settings I'm describing.

I'm attaching the xorg logs and the config.

I user latest mesa/drm and radeon driver in kms mode, with kernel 2.6.32.
Comment 1 Aljaž Prusnik 2009-12-05 05:11:55 UTC
Created attachment 31764 [details] [review]
no additional settings in xorg (just radeon driver under driver section)
Comment 2 Aljaž Prusnik 2009-12-05 05:12:48 UTC
Created attachment 31765 [details] [review]
This is the log when I only attach on a HDMI connector (correct behaviour)
Comment 3 Aljaž Prusnik 2009-12-05 05:13:41 UTC
Created attachment 31766 [details] [review]
this is when all other options in xorg are uncommented (both monitor and screen sections)
Comment 4 Aljaž Prusnik 2009-12-05 05:15:10 UTC
Created attachment 31767 [details] [review]
Xorg configuration file for the default xorg.log
Comment 5 Alex Deucher 2009-12-05 09:05:54 UTC
Can you attach your dmesg output?
Comment 6 Aljaž Prusnik 2009-12-05 11:31:25 UTC
Created attachment 31768 [details] [review]
dmesg, default xorg .conf (only radeon driver)

it seems that it recognizes proper resolutions but cannot figure out why it does not apply them.
With this default setting I get 1600x1200 resolution on both monitors (and not 1920x1200 and 1600x1200).
Comment 7 Alex Deucher 2009-12-05 15:35:32 UTC
Does non-kms work properly?  (load the radeon drm with modeset=0)
Comment 8 Alex Deucher 2009-12-05 15:38:41 UTC
Does setting up dualhead using xrandr from the command line work?  The gui randr tools tend to be problematic.  E.g., start X and run:
xrandr --output DVI-0 --right-of HDMI-0
Comment 9 Aljaž Prusnik 2009-12-05 15:49:45 UTC
Without the modesetting it seems to be working properly. I'll attach the xorg.log after succesfully establishing the dual head setup.

Regarding the xrandr - no it did not help. The behaviour was the same as when I used Gnome display settings.

Comment 10 Aljaž Prusnik 2009-12-05 15:53:33 UTC
Created attachment 31771 [details]
xorg log radeon.modesetting=0, dualhead enabled

Dual head was enabled with gnome display settings.

How it works now
Default setting was Mirror. After I unchecked the mirror, I had to set the virtual size in xorg.log, so I uncommented those lines. Then I could set both screens properly, with each one being labeled by their names (HP and Philips, respectively).

How it works under KMS
The same exercise, only on both screens it said Philips and both displays were behaving as "right of".
Comment 11 Aljaž Prusnik 2009-12-13 09:23:37 UTC
I just updated all the mesa/drm/radeon thing and it's still the same. Kernel log says this when doing the switching between monitors (interesting enough: if I turn off DVI-0 display, the HDMI display turns off as well. It's as though DVI and HDMI were treated as though it's the same output):

Dec 13 18:05:15 slash kernel: [   50.879331] Unpin not necessary for ffff88011c443e00 !
Dec 13 18:08:11 slash kernel: [  227.288279] executing set pll
Dec 13 18:08:11 slash kernel: [  227.293019] executing set crtc timing
Dec 13 18:08:11 slash kernel: [  227.293052] [drm] TMDS-15: set mode  39
Dec 13 18:08:17 slash kernel: [  232.922805] Unpin not necessary for ffff880037b91000 !
Dec 13 18:08:18 slash kernel: [  234.322404] executing set pll
Dec 13 18:08:18 slash kernel: [  234.322418] executing set crtc timing
Dec 13 18:08:18 slash kernel: [  234.322450] [drm] TMDS-15: set mode  41
Dec 13 18:08:18 slash kernel: [  234.355414] executing set pll
Dec 13 18:08:18 slash kernel: [  234.360145] executing set crtc timing
Dec 13 18:08:18 slash kernel: [  234.360176] [drm] TMDS-11: set mode  44
Dec 13 18:08:49 slash kernel: [  265.250608] executing set pll
Dec 13 18:08:49 slash kernel: [  265.250631] executing set crtc timing
Dec 13 18:08:49 slash kernel: [  265.250677] [drm] TMDS-15: set mode 1600x1200 38
Dec 13 18:09:15 slash kernel: [  290.625331] Unpin not necessary for ffff880037a78a00 !
Dec 13 18:09:15 slash kernel: [  291.313555] Unpin not necessary for ffff880116e71200 !
Dec 13 18:09:25 slash kernel: [  300.455526] Unpin not necessary for ffff880116e71c00 !
Dec 13 18:10:08 slash kernel: [  343.951414] executing set pll
Dec 13 18:10:08 slash kernel: [  343.956147] executing set crtc timing
Dec 13 18:10:08 slash kernel: [  343.956188] [drm] TMDS-15: set mode  42
Dec 13 18:11:13 slash kernel: [  408.437287] Unpin not necessary for ffff8801068d5800 !
Dec 13 18:11:14 slash kernel: [  410.154389] executing set pll
Dec 13 18:11:14 slash kernel: [  410.159148] executing set crtc timing
Dec 13 18:11:14 slash kernel: [  410.159187] [drm] TMDS-11: set mode  4d
Dec 13 18:11:14 slash kernel: [  410.192428] executing set pll
Dec 13 18:11:14 slash kernel: [  410.192447] executing set crtc timing
Dec 13 18:11:14 slash kernel: [  410.192485] [drm] TMDS-15: set mode  50
Dec 13 18:12:30 slash kernel: [  485.743414] executing set pll
Dec 13 18:12:30 slash kernel: [  485.748148] executing set crtc timing
Dec 13 18:12:30 slash kernel: [  485.748189] [drm] TMDS-15: set mode  4e
Dec 13 18:13:40 slash kernel: [  555.499884] Unpin not necessary for ffff880106bcdc00 !
Dec 13 18:13:41 slash kernel: [  556.795264] executing set pll
Dec 13 18:13:41 slash kernel: [  556.800023] executing set crtc timing
Dec 13 18:13:41 slash kernel: [  556.800062] [drm] TMDS-11: set mode  57
Dec 13 18:13:41 slash kernel: [  556.833305] executing set pll
Dec 13 18:13:41 slash kernel: [  556.833325] executing set crtc timing
Dec 13 18:13:41 slash kernel: [  556.833363] [drm] TMDS-15: set mode  5a
Dec 13 18:18:14 slash kernel: [  829.987276] Unpin not necessary for ffff8800de043000 !
Dec 13 18:18:15 slash kernel: [  831.045306] executing set pll
Dec 13 18:18:15 slash kernel: [  831.045326] executing set crtc timing
Dec 13 18:18:15 slash kernel: [  831.045366] [drm] TMDS-15: set mode  61
Dec 13 18:18:42 slash kernel: [  857.858290] executing set pll
Dec 13 18:18:42 slash kernel: [  857.863022] executing set crtc timing
Dec 13 18:18:42 slash kernel: [  857.863063] [drm] TMDS-15: set mode  58
Comment 12 Aljaž Prusnik 2009-12-22 16:58:43 UTC
Under 2.6.33-rc1 and drm-next it seems to work properly (with the helo of friends on IRC). Should this bug be closed or should it wait until the stable release?
Comment 13 Aljaž Prusnik 2009-12-25 09:48:55 UTC
I just compiled the new 2.6.33-rc2 and it works under this kernel, so I'll take the liberty and close this bug.


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.