Bug 29041 - [MMIO] column ordering (23endianness?) issue with video overlay
Summary: [MMIO] column ordering (23endianness?) issue with video overlay
Status: RESOLVED FIXED
Alias: None
Product: xorg
Classification: Unclassified
Component: Driver/Radeon (show other bugs)
Version: 7.5 (2009.10)
Hardware: PowerPC Linux (All)
: medium normal
Assignee: xf86-video-ati maintainers
QA Contact: Xorg Project Team
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2010-07-13 06:03 UTC by FatPhil
Modified: 2010-10-19 15:27 UTC (History)
0 users

See Also:
i915 platform:
i915 features:


Attachments

Description FatPhil 2010-07-13 06:03:02 UTC
Forgive the newb-ness of this report - I've been running linux/ppc64 on this dual G5 powermac desktop machine for no more than a day, and barely even know what hardware it's got inside it.


phil@geespaz:~$ dmesg | grep -i radeon 
[    0.546025] radeonfb 0000:f0:10.0: enabling device (0006 -> 0007)
[    0.744230] radeonfb: Found Open Firmware ROM Image
[    0.744241] radeonfb: Retrieved PLL infos from Open Firmware
[    0.744249] radeonfb: Reference=27.00 MHz (RefDiv=12) Memory=200.00 Mhz, System=325.00 MHz
[    0.744256] radeonfb: PLL min 12000 max 35000
[    1.502451] radeonfb: Monitor 1 type DFP found
[    1.502456] radeonfb: EDID probed
[    1.502461] radeonfb: Monitor 2 type no found
[    1.579040] radeonfb (0000:f0:10.0): ATI Radeon 4150 "AP"
[  703.929179] [drm] radeon kernel modesetting enabled.

OS-wise, this is
Comment 1 Michel Dänzer 2010-07-13 06:08:38 UTC
(In reply to comment #1)
> [    1.579040] radeonfb (0000:f0:10.0): ATI Radeon 4150 "AP"
> [  703.929179] [drm] radeon kernel modesetting enabled.

radeonfb and kernel modesetting (KMS) are mutually exclusive, and attempting to enable the latter when the former is active currently results in the radeon kernel module not initializing at all, preventing the DRI from being enabled. If you either disable radeonfb or pass radeon.modeset=0 on the kernel command line, video should work fine again.
Comment 2 FatPhil 2010-07-13 06:26:21 UTC
Mid-air collision!!!

I'll try what you suggest, and report back. Until then, here's the rest of what I was going to write...

F*cking synaptics touchpad on my laptop...

OS-wise this is 

phil@geespaz:~$ cat /etc/debian_version 
squeeze/sid
phil@geespaz:~$ uname -a
Linux geespaz 2.6.32-5-powerpc64 #1 SMP Tue Jun 1 15:09:03 UTC 2010 ppc64 GNU/Linux

phil@geespaz:~$ dpkg -l | grep radeon
ii  libdrm-radeon1                          2.4.18-6                       Userspace interface to radeon-specific kerne
ii  radeontool                              1.6.1-1                        utility to control ATI Radeon backlight func
ii  xserver-xorg-video-radeon               1:6.13.0-2                     X.Org X server -- AMD/ATI Radeon display dri

phil@geespaz:~$ dpkg -l | grep xorg  
ii  xorg                                    1:7.5+6                        X.Org X Window System
ii  xorg-docs-core                          1:1.5-1                        Core documentation for the X.org X Window Sy
ii  xserver-xorg                            1:7.5+6                        the X.Org X server
ii  xserver-xorg-core                       2:1.7.7-2                      Xorg X server - core server
...


The problem is best illustrated with an image, but as it's the overlay, it has to be a photo rather than a screengrab. Here's a camphone snap of it
  http://fatphil.org/images/XoverlayPPC64_1.jpg
(which I guess I should attach to this ticket).

The issue takes a couple of different forms depending on precisely which video format and player I use. One AVI always (VLC, xine, and mplayer) has odd and even columns swapped over, giving horrid sawtooth jaggies down all sloping edges. Another AVI is like that with mplayer and xine, but in vlc the columns are in the right order, but the colours are completely messed up. If I use mplayer -vo x11 or mplayer -vo gl, then the columns are all in the right place, it's only -xo xv that shows that particular issue.
Comment 3 FatPhil 2010-07-13 07:23:57 UTC
Booting with video=radeonfb:off
No consoles, no X. A bit of a dead end presently.

Booting with radeon.modeset=0
I get to X OK.
vlc causes the screen to go blank, killing it doesn't help, I needed to restart X.
xine displays its logo and plays a short video correctly - I'll test this more.
mplayer plays a few frames of the video correctly, but then the screen went black and my SSH session died, and the machine needed a reboot.

The xine result implies that Michel's diagnosis was an accurate one. I only need one video player, so if it can play whole movies this might be enough.
Comment 4 Michel Dänzer 2010-07-13 07:38:39 UTC
(In reply to comment #3)
These are all separate issues that would need to be tracked separately, but FWIW...

> Booting with video=radeonfb:off
> No consoles, no X. A bit of a dead end presently.

At the very least the kernel output, but preferably also the Xorg.0.log file would be needed to diagnose this.


> Booting with radeon.modeset=0
> I get to X OK.
> vlc causes the screen to go blank, killing it doesn't help, I needed to restart
> X.

This could be a number of things, e.g. screensaver / DPMS / gamma (does running xgamma -gamma 1.0 restore the normal display?) related.


> mplayer plays a few frames of the video correctly, but then the screen went
> black and my SSH session died, and the machine needed a reboot.

Stability issues like that are usually AGP related, you could try playing with all the possible values for Option "AGPMode" (see the radeon manpage) or even disabling AGP altogether with Option "BusType" "PCI".
Comment 5 FatPhil 2010-07-20 02:48:20 UTC
Michel's solution works for me, and has no noticeable negative side effects.
Many thanks! FIXED...
Comment 6 FatPhil 2010-07-20 02:48:42 UTC
... and VERIFIED.
Comment 7 Michel Dänzer 2010-07-20 03:05:07 UTC
A workaround is a workaround, even if it's actually desirable in this case. :) The bug is still there.
Comment 8 Alex Deucher 2010-10-19 15:27:42 UTC
fixed in:
91f707d308d4bbf16c3d62d046cf280fef5a8f4c


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.