(Needs radeon-rewrite) 3D stuff breaks with Compiz: Redirected Direct Rendering is needed in DRI

[Alexander Jones]

If I run glxgears or anything that uses 3D while running Compiz, it's pretty obvious that the graphics hardware is just rendering straight on top of the screen rather than to a window.

In windowed mode, the rendering stays always on top, and moving a window causes a stale rendering to be left behind.

In fullscreen mode, windows underneath cause a flickering effect.

This is on R350/radeon hardware.

ProblemType: Bug
Architecture: i386
Date: Tue Mar 27 15:57:27 2007
DistroRelease: Ubuntu 7.04
Uname: Linux flash 2.6.20-13-generic #2 SMP Sun Mar 25 00:21:25 UTC 2007 i686 GNU/Linux

-----
Issues with Composite and 3D are widely reported and show themselves in a variety of contexts, but all are derived from the same root problem. 3D apps wish to do direct rendering in order to maximize performance, but Compiz wishes to retain control over how all windows are painted and doesn't work well when apps are rendering directly. Thus, what's needed is a method to redirect these directly rendering apps. Redirected direct rendering is nicely described in this blog post:

http://hoegsberg.blogspot.com/2007/08/redirected-direct-rendering.html

The general effect is that when Compiz is active, running a 3D app like glxgears will show rendering errors when moving the glxgears window or putting it behind or over other windows. The standard workaround is to disable compiz while using the 3D programs and re-enable it when done.

This is a general problem in X's core architecture, and not an issue particular to a given driver. It's not exactly Compiz's fault either, and unfortunately it lacks a mechanism to hide or work around the problem. Solving it requires modification of a number of components including adding the TTM memory manager to libdrm, modifications of DRI in X, and on and on. It represents a significant amount of work to be done upstream and will probably take on the order of a year or two to get everything sorted out.

[Joe Mulloy (jdmulloy)]

Created an attachment (id=7494)
Screenshot fo problem

This screenshot shows glxgears running behind Konsole with xcompmgr running,
but this happens with other composite managers as well and with other OpenGL
programs such as screensavers.

[Joe Mulloy (jdmulloy)]

Created an attachment (id=7495)
Xorg Configuration File

[Joe Mulloy (jdmulloy)]

Created an attachment (id=7496)
Xorg log

[Alexander Jones (alex-weej)]

If I run glxgears or anything that uses 3D while running Compiz, it's pretty obvious that the graphics hardware is just rendering straight on top of the screen rather than to a window.

In windowed mode, the rendering stays always on top, and moving a window causes a stale rendering to be left behind.

In fullscreen mode, windows underneath cause a flickering effect.

This is on R350/radeon hardware.

ProblemType: Bug
Architecture: i386
Date: Tue Mar 27 15:57:27 2007
DistroRelease: Ubuntu 7.04
Uname: Linux flash 2.6.20-13-generic #2 SMP Sun Mar 25 00:21:25 UTC 2007 i686 GNU/Linux

[Alexander Jones (alex-weej)]

This is still affecting 7.04.

[Michel-tungstengraphics (michel-tungstengraphics)]

Fixing this is far from trivial, see http://dri.freedesktop.org/wiki/DirectRenderingToRedirectedWindows for a sketch of the road ahead.

[Joe Mulloy (jdmulloy)]

Why does EVERYONE ELSE with this EXACT chip not have this problem. Is it actually a driver problem? I've read reports of people being able to run Beryl with no problem on this chip. How do I prove this is a Hardware fault so Toshiba will fix it?

[Michel-tungstengraphics (michel-tungstengraphics)]

(In reply to comment #5)
> Why does EVERYONE ELSE with this EXACT chip not have this problem.

How did you get that impression? Everybody should have the same problem under the same circumstances.

> Is it actually a driver problem?

It's more complicated than that, see comment #4.

> I've read reports of people being able to run Beryl with no problem on this
> chip.

I can only guess based on this little information, but maybe they're using Xgl.

> How do I prove this is a Hardware fault

It probably isn't.

[Alexander Jones (alex-weej)]

It seems that indirect rendering is failing.

[Bryce Harrington (bryce)]

Hi Alex, thanks for reporting this bug.

Can you attach your /etc/X11/xorg.conf and /var/log/Xorg.0.log files?

[Gigaplex (gigaplex)]

I have the same problem on my laptop (incidentally also a Toshiba M100). I never noticed it until I stumbled across this bug report and checked for myself.

[Russ-csh (russ-csh)]

Is there any work currently to have it fixed? Or is the fix still being merely discussed...

[Kristian Hoegsberg (krh-bitplanet)]

(In reply to comment #8)
> Is there any work currently to have it fixed? Or is the fix still being merely
> discussed...

There is work going on here, but the scope of this is a lot bigger than your average bug fix. We need to rework the interaction and interface between the kernel drm, the X server and the 3D DRI drivers. We're probably talking man-years of work here, so you'll have to wait a little longer :) Michels roadmap mentioned in comment 4 gives an idea of the changes we're looking at.

[AJ (s800human)]

Subject: 3D Windows plugin in Compiz Fusion:
Problem: Windows are breaking when they are in more than 1 viewport

This plugin was working perfectly in Beryl but in Compiz Fusion I having a problem as shown in screenshot.

I know that many of the things are in progress but just for your information

[Alexander Jones (alex-weej)]

s800human, this is not the same bug. Please file it elsewhere! Thanks.

[AJ (s800human)]

I forgot to mention that this problem arise when the window is lifted above from cube using this 3D plugin.

[AJ (s800human)]

Ok. Sorry Alex. I just read the title.

[Alexander Jones (alex-weej)]

Attached

[Michel-tungstengraphics (michel-tungstengraphics)]

*** Bug 6283 has been marked as a duplicate of this bug. ***

[Michel-tungstengraphics (michel-tungstengraphics)]

*** Bug 1208 has been marked as a duplicate of this bug. ***

[Nicolò Chieffo (yelo3)]

I'm in gusty now, with ati 6.7.194 and I have the same problem of the screenshot

a similar (but not identic) thing happens when playing videos: when I move the window, the video becomes black (and when activating the scale plugin the window is black too, and returns with content only after having finished the scale animation)

glxgears + scale instead leaves the 3d output where the window was

[Joe Mulloy (jdmulloy)]

Any progress? This bug is over a year old. I bought a laptop with an intel graphics chip because I wanted to use a free software driver but obviously that was a mistake. Next time I'll go with an NVIDIA chip. The drivers may be proprietary but at least their stuff WORKS! It's nearly 2008, I should be able to have nice effects on my computer. KDE 4 is coming out soon and I won't be able to use it to it's full potential because this is BROKEN!

[Daniel Stone (daniels)]

It's your decision to make: don't let us stop you.

[Jesse Barnes (jbarnes-virtuousgeek)]

Joseph, there's been lots of progress on this front, but the final bits haven't been released yet afaik, though Kristian may have more details on what's available for testing etc.

[Olig9 (olig9)]

Isn't there any kind of workaround for this? I don't care much for performance on the particular machine where this problem occurs, but it would be nice if toys like Google Earth or Celestia would have usable menus. Any idea?

[Michel-tungstengraphics (michel-tungstengraphics)]

(In reply to comment #15)
> Isn't there any kind of workaround for this?

Unfortunately not - the only way to get correct compositing is to properly redirect the rendering.

[Michel-tungstengraphics (michel-tungstengraphics)]

*** Bug 14034 has been marked as a duplicate of this bug. ***

[e (0x0065)]

I would be interested in amd64 testing, when something's available.

I'm having the same problem on my desktop.
Q965 Integrated Graphics Controller (rev 02) --> Intel® GMA 3000

It will likely affect my eeepc also.

thx

e

[Timo Aaltonen (tjaalton)]

This is a general issue with the current infrastructure. Upstream is working on this, so maybe for Hardy+1 or 2 we'll have working redirected direct rendering in place.

[Sven Lamers (mrfuji)]

After installing Hardy Beta on a friend's PC this suddenly works without a flaw, but he's using the "new card" nvidia driver. However, when I tried it at home on my Intel GMA945 laptop I still had this bug.

Of course, this does not only affect glxgears but also more productive apps like stellarium and celestia who aren't usable at all.
So my question is, IF this is a problem with the current infrastructure, why does it work with the nvidia driver? Makes me think it's more driver related....

[Timo Aaltonen (tjaalton)]

This bug affects the opensource drivers.

[Sven Lamers (mrfuji)]

Uh, sure... Ok, my bad. Sorry :)

[Alexander Jones (alex-weej)]

Indeed, the only setup for which this works is with NVidia cards and
their proprietary driver. (I am still astonished this is default
behaviour.)

[Sofa (alexschultze)]

The same here. I can upload a vid for everyone who cannot imagine it yet :-)

[Oliver Mattos (omattos)]

On some systems this "bug" causes flickering as both compiz and the 3d app are writing to the same bit of framebuffer even when there are no 3d effects going on, making the 3d app unusable. This also occurs in fullscreen mode where there is really no need for compiz to be touching the framebuffer at all.

The most visible effect of this is that most of the included screensavers flicker, the included 3d games are unusable, and apps like google earth or opengl toolkits are pretty hard to use.

Since this bug looks like it'll take quite a few releases to fix, could a little app be written to detect that a 3d app is running and turn composite off. when the 3d app quits it can restart composite? (thats what Windows does for legacy apps that communicate directly with the graphics hardware)

[Alexander Jones (alex-weej)]

Oliver, I'm not sure it's that bad -- "Un-direct fullscreen windows" should be enabled by default. i.e. full screen windows should draw directly to the framebuffer. Check your Compiz configuration if you've messed with it.

My advice, however, is if your driver doesn't support indirect 3D (that is, any driver other than proprietary nvidia), disable Compiz. It's really not worth the grief.

[Oliver Mattos (omattos)]

yep, but remember a good 30% of users use non nvidia cards. This needs to
be detected more intelligently. Maybe compiz should not be enabled by
default for non-nvidia cards until this problem is fixed.

On Thu, Apr 24, 2008 at 8:01 PM, Alexander Jones <email address hidden> wrote:

> Oliver, I'm not sure it's that bad -- "Un-direct fullscreen windows"
> should be enabled by default. i.e. full screen windows should draw
> directly to the framebuffer. Check your Compiz configuration if you've
> messed with it.
>
> My advice, however, is if your driver doesn't support indirect 3D (that
> is, any driver other than proprietary nvidia), disable Compiz. It's
> really not worth the grief.
>
> --
> 3D stuff breaks with Compiz: Redirected Direct Rendering is needed in DRI
> https://bugs.launchpad.net/bugs/96991
> You received this bug notification because you are a direct subscriber
> of the bug.
>

[Bryce Harrington (bryce)]

Setting to alpha-5 for intrepid, as we tentatively expect this feature to become available by the time intrepid is released.

[Neil J. Patel (njpatel)]

Sorry for the spam, but is there any update on this?

Although I understand that this is not, by any stretch of the imagination, a trivial fix, there seem to be more and more Linux-based devices that are coming out in the next few months that all exhibit this issue, and it sucks to have to choose between composite or GL. This suckage increases as we are now starting to get more application<=>GL integration on the desktop.

Anyway, thanks for all the hard work!

PS. I can provide testing on the 945GSE chipset.

[Joe Mulloy (jdmulloy)]

There should be a fix for this in the next release.

It's called DRI2.

See
http://hoegsberg.blogspot.com/2008/03/i-just-committed-last-bit-of-dri2-work.html
http://www.x.org/wiki/DRI2
http://lists.freedesktop.org/archives/xorg/2008-February/032646.html

Anyone know how I can get a version of Xorg with DRI2 on Gentoo?

[saads (shakhshir)]

Does this bug explain the bug in https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/185370, which has been marked as a dup of this one? I am facing the problem found in that bug and would like to confirm, otherwise I would like to unmark it as a dup and reopen the bug.

[Oliver Mattos (omattos)]

Yes - it is based on the same issue - there is no proper memory management
for graphics memory, and therefore two instances of Compiz can't get on
together.

On Fri, May 30, 2008 at 7:40 AM, saads <email address hidden> wrote:

> Does this bug explain the bug in
> https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-
> intel/+bug/185370<https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/185370>,
> which has been marked as a dup of this one? I am
> facing the problem found in that bug and would like to confirm,
> otherwise I would like to unmark it as a dup and reopen the bug.
>
> --
> 3D stuff breaks with Compiz: Redirected Direct Rendering is needed in DRI
> https://bugs.launchpad.net/bugs/96991
> You received this bug notification because you are a direct subscriber
> of the bug.
>

[Daniel Stone (daniels)]

Hooray for DRI2!

[Patricio (patricio-stwing)]

What is the status? I see the last post was in May. I have the same problem with a scientific software that wants to render 3D images.
thanks!
Patricio

[Ivo Roghair (ivo-82)]

Same for me. My computer runs desktop effects on an intel X3100 (GM965) with AIGLX (automatically detected flawlessly by default), however running openGL applications, I cannot use desktop effects because of the flickering. However, if I install xserver-xgl everything works fine, except that full screen videos do not play nicely (they are slowish, and jerky). The full screen video does play well under xgl when I right-click the movie and the popup menu shows... Also, on xgl, opengl graphics are much slower than on aiglx.

So, I'll install xserver-xgl for now, but running the opengl apps on aiglx would improve my systems performance drastically.

[Bryce Harrington (bryce)]

Status update:
Unfortunately things do not look good. Upstream has changed their mind, and instead of pursuing the TTM memory manager (which turned out to be too complex) are instead going to develop a new memory manager called GEM. The timeframe for seeing GEM released for our use doesn't meet the Intrepid schedule. We're hoping to see it for Intrepid+1, but we will see.

Please see the following article and its associated links for more details:
http://www.phoronix.com/scan.php?page=news_item&px=NjUyMQ

[Nicolò Chieffo (yelo3)]

Does this mean that we won't have compiz in 8.10?

[Alexander Jones (alex-weej)]

No, it means the issues it introduces with anything other than the
NVidia proprietary driver are just not fixable until then.

[Oliver Mattos (omattos)]

Could I put forward the idea of disabling compiz by default on non- nvidia
cards for the intrepid release, since from my perspective making all 3d apps
unusable on this hardware by default is unsatisfactory.

Either that or a tray icon to disable compiz.

Or even better automaticly disable compiz when an opengl app attempts to
make a drawing surface.

On Fri, Jul 18, 2008 at 6:16 PM, Alexander Jones <email address hidden> wrote:

> No, it means the issues it introduces with anything other than the
> NVidia proprietary driver are just not fixable until then.
>
> --
> 3D stuff breaks with Compiz: Redirected Direct Rendering is needed in DRI
> https://bugs.launchpad.net/bugs/96991
> You received this bug notification because you are a direct subscriber
> of the bug.
>

[Nicolò Chieffo (yelo3)]

Oh, do you mean no compiz for Intel video card owners?
This is a huge regression! Can't the original driver be used instead
of the new one?

[Alexander Jones (alex-weej)]

The NVidia proprietary driver rips out a lot (the whole?) of the
device-dependent X code and replaces it with its own. For this reason,
it is the only system that has the ability to accelerate indirect 3D
drawing and allow multiple users of Compiz at the same time. The open
source drivers are still trying to find their way.

Disabling one of our flagship eye candy pieces is almost an admission
of defeat, and while I agree with the practical argument, I don't see
this being a popular choice amongst "the masses".

[Ivo Roghair (ivo-82)]

Also in the current beta/liveCD from intrepid.

glxgears does display when compiz is enabled, however dragging the window shows artifacts (the gears stay at their original positions until the button for window dragging is released, then the rotating gears are shown at the correct location, however, a single frame will still be shown at the old position of the window. Also if another window is on top of glxgears, the gears are still shown (whereas the window border isnt) -- intel x3100 laptop

[Vytas (vytas)]

I think this is a serious bug and Compiz should be turned off by default on all cards that do not fully support compositing. Users who can somehow live with this problem can always switch it on manually.

Yes, such decision would be called retreat, defeat or w/e but IMHO enabling obviously crippled software does more harm than good. Sooner or later users run into this problem, and go report it as a bug, flooding the bug system, or even worse, conclude that Ubuntu is alpha quality.

[Alexander Jones (alex-weej)]

I have made this argument for years. It is falling on deaf ears.

[David Fokkema (dfokkema)]

Second that. That is, please disable compiz on not well supported cards.
(Or, at least, mention that 3D support with compiz is broken for some
cards since on intel, 2D support is fine and compiz is very nice).

[DelSol (home-snkplay)]

i dont understand how this bug in compiz can survive so long. so many people know and complain about it, so why is nothing happening? this bug is in compiz for a long time, so turn it off if it doesnt work right. cmon. make the system STABLE and dont use things that contain bugs.
even in intrepid beta you get the graphical problems - thats just a pain in the ass....

[Zaca (zaca98)]

This guy has apparently solved the problem:

http://hoegsberg.blogspot.com/2008/03/i-just-committed-last-bit-of-dri2-work.html

However, it is not clear if, or when, his solution will be ported to the main distributions. It is clear that it was not delivered on Intrepid. Maybe on Fedora 10? Anyone has made tests on that?

[Tormod Volden (tormodvolden)]

You will might see DRI2 in Ubuntu 9.04. It's just now coming to the upstream main trees and releases.

[Bryce Harrington (bryce)]

Jaunty Alpha-2 now includes DRI2, and -intel and -ati should support it properly now. Can someone please confirm that this issue is no longer occurring?

[Brian Rogers (brian-rogers)]

The necessary work is in the X server 1.6 branch. It's all up to the drivers now. DRI2 code is in the packaged Intel driver, but not enabled by default yet.

[Robert Persson (ireneshusband)]

I don't understand why this bug is "wishlist". It's broken isn't it?

[Ivan Stetsenko (stetzen)]

I'm using -intel driver on Jaunty, and DRI2 doesn't work for now (at least, 3d apps like glxgears are still broken). I hope, it'll be fixed quit soon.

[Michel Dänzer (michel-daenzer)]

*** Bug 19742 has been marked as a duplicate of this bug. ***

[Saivann Carignan (oxmosys)]

Using -intel, -ati and -nvidia (proprietary) on jaunty alpha 4, I got no issues running glxgears or google earth with any compiz plugin (wobbly, transparency, expo) and I am able to put windows over Google Earth as much as glxgears without getting weird behaviors. So far, all my tests seems to reveal that this bug is fixed.

[Roberto Cássio Jr. (rcsdnj)]

Saïvann,

     Did you have to enable UXA acceleration (Option "AccelMethod" "UXA" on
xorg.conf) on -intel drivers to get redirected direct rendering work? I had
to. Is this the expected behavior?

On Sun, Feb 8, 2009 at 2:08 AM, Saïvann Carignan <email address hidden> wrote:

> Using -intel, -ati and -nvidia (proprietary) on jaunty alpha 4, I got no
> issues running glxgears or google earth with any compiz plugin (wobbly,
> transparency, expo) and I am able to put windows over Google Earth as
> much as glxgears without getting weird behaviors. So far, all my tests
> seems to reveal that this bug is fixed.
>
> --
> 3D stuff breaks with Compiz: Redirected Direct Rendering is needed in DRI
> https://bugs.launchpad.net/bugs/96991
> You received this bug notification because you are a direct subscriber
> of the bug.
>

[Tormod Volden (tormodvolden)]

I do not get it to work on -ati (on X700 card):
[ 0.275470] (II) Loading extension DRI2
[ 1.833171] (II) AIGLX: Screen 0 is not DRI2 capable
and when I run glxgears it does not follow the rotating cube.

[Saivann Carignan (oxmosys)]

Roberto and Tormod : I re-verified with today updates and you're right, it does not work.. I don't know why but it worked a few days ago with both ati and intel (at least for me).. It still works perfectly with nvidia though but not open source ati and intel drivers. Shouldn't we open a bug status for intel and ati packages?

[Tormod Volden (tormodvolden)]

Well, nvidia is a special case since it ships its own proprietary libraries and shortcuts the normal X stack. So unless we know something is needed in the ati or intel drivers and not in the normal Xorg and mesa stuff, I think we should leave the bug status as it is.

[suoko (suoko)]

great. i confirm jaunty solved the compiz bug with 3d apps on intel cards

2009/2/9 Tormod Volden <email address hidden>

> Well, nvidia is a special case since it ships its own proprietary
> libraries and shortcuts the normal X stack. So unless we know something
> is needed in the ati or intel drivers and not in the normal Xorg and
> mesa stuff, I think we should leave the bug status as it is.
>
> --
> 3D stuff breaks with Compiz: Redirected Direct Rendering is needed in DRI
> https://bugs.launchpad.net/bugs/96991
> You received this bug notification because you are a direct subscriber
> of a duplicate bug.
>

--
Gabriele Vidali

[Eric Anholt (eric-anholt)]

*** Bug 19974 has been marked as a duplicate of this bug. ***

[Roberto Cássio Jr. (rcsdnj)]

Hi,
   On my Jaunty installation, it is still required to set "Option
AccelMethod" "UXA"" in xorg.conf to make DRI2 and redirected direct
rendering work. I'm using Intel drivers. Considering that UXA is not enabled
by default and DRI2 doesn't work on EXA, should this really be mark as "Fix
Released"?

On Fri, Feb 13, 2009 at 12:19 PM, Davidiam <email address hidden> wrote:

> ** Changed in: xorg-server (Ubuntu)
> Status: Triaged => Fix Released
>
> --
> 3D stuff breaks with Compiz: Redirected Direct Rendering is needed in DRI
> https://bugs.launchpad.net/bugs/96991
> You received this bug notification because you are a direct subscriber
> of the bug.
>

[Andy Clayton (q3aiml)]

Correct me if I am wrong, but isn't it a bit premature to be marking this as Fix Released? It only works on -intel if you edit your xorg.conf and enable UXA. That's hardly a fix though, as it's quite unstable for a number of people and not likely to be the default in jaunty (https://wiki.ubuntu.com/X/UxaTesting) (https://lists.ubuntu.com/archives/ubuntu-x/2009-February/000391.html).

[Brian Rogers (brian-rogers)]

The server implements what it needs to now, so "fix released" is appropriate. However, we're still waiting for the driver to turn on DRI2 by default.

[Robert Persson (ireneshusband)]

What do you have to put into /etc/hal/fdi/policy to get the equivalent of "option accelmethod uxa"?

[Saivann Carignan (oxmosys)]

According to developers blogs, both intel and ati drivers have a lot of work done in that direction but does not enable it by default until it will be stable enough. It's only a matter of time!

http://keithp.com/blogs/UMA_Acceleration_Architecture/
http://jglisse.livejournal.com/1623.html

[Bryce Harrington (bryce)]

Correct; from xserver's perspective this is now a done deal. It is a matter of each driver now adding support for it and setting it on by default.

-intel supports it now but not by default. To get it, you must specify Option "AccelMethod" "UXA" in your xorg.conf. UXA is not yet stable enough to switch on by default but hopefully should be by Karmic.

-ati and -nouveau have further to go before we can consider it supported there.

-nvidia sort of does its own thing and probably will never support DRI2.

-fglrx is anyone's guess, but if it does plan to support it, it'll probably be a long while. We're better off focusing on gaining support for it in -ati.

For other drivers, I don't have a clear idea what the plans are. Less well maintained video drivers probably will never gain support. But feel free to add bug tasks for drivers you think need to get DRI2 that we should be tracking.

[Nicolò Chieffo (yelo3)]

Yes, I'm in UXA and I can play 3d games with no more problmes

[Timo Aaltonen (tjaalton)]

nouveau is still quite immature, and 3D is completely unsupported, so lowering importance due to that.

[Chris Halse Rogers (raof)]

 affects ubuntu/xserver-xorg-video-nouveau
 status wontfix

The nouveau drivers in Ubuntu offer absolutely no 3d support; this bug
doesn't really apply to them at the moment.

The upstream nouveau development does have some 3d support for some
cards, but this is aggressively unsupported - it is known to be
incomplete, and the best that can be expected is for it to run
undemanding 3d apps until VRAM is exhausted. Upstream does not want to
hear about problems with 3D unless there is a patch attached which fixes
them!

By the time nouveau's 3D support is worth shipping, it will support DRI2
- the initial work is already there.

[Martin Schaaf (mascha)]

I can confirm that this is fixed int ubuntu jaunty alpha5 by enabling dri2 and uxa. Thank you very much.

[Martin Schaaf (mascha)]

This is fixed for the intel driver int jaunty by enabling dri2 and uxa.

[Julien Olivier (julo)]

Marti Schaaf wrote: "This is fixed for the intel driver int jaunty by enabling dri2 and uxa."

I don't have a xorf.conf file, so I guess worg is auto-configured. If dri2 and uxa are needed to fix this bug, would it be made so that they are activated by default when not using any xorg.conf ?

Also, I'd like to test it, so how can I enable dri2 and uxa easily ?

Thanks for your help.

[David E. Fox (dfox94085)]

On Wed, May 6, 2009 at 12:56 PM, Julien Olivier <email address hidden> wrote:

> Also, I'd like to test it, so how can I enable dri2 and uxa easily ?

Easiest way is to add

 Option "AccelMethod" "UXA"

in your Device section in the /etc/X11/xorg.conf file.

--
thanks for letting me change the magnetic patterns on your hard disk.

[Mackenzie Morgan (maco.m)]

Create an xorg.conf
sudo dexconf -o /etc/X11/xorg.conf
and then edit that.

And no, they are not enabled by default because they're experimental and on
some Intel graphics chips they make X crash.

[Julien Olivier (julo)]

Thanks for the information Mckenzie and David.

So, I managed to enable UXA and DRI2 and now OpenGL apps work flawlessly (even if way slower) in fullscreen mode with compiz enabled.

However, the bug with videos disappearing when moved/resized is still there.

[Mackenzie Morgan (maco.m)]

Can you add the bit about it being slower to
http://wiki.ubuntu.com/X/UxaTesting ? It's supposed to be faster...

[Pablo Estigarribia (pablodav)]

Hi t fix all these things I updated xserver-xorg-video-intel to version 2.7 and put this in /etc/X11/xorg.conf:
Section "Device"
 Identifier "Configured Video Device"
 Option "UseFBDev" "true"
 Option "AccelMethod" "UXA"
EndSection

(It could be done with sudo gedit /etc/X11/xorg.conf)

Follow this page to get instructions how to install latest version of intel driver from x-updates: https://launchpad.net/~ubuntu-x-swat/+archive/x-updates

The latest version of intel video driver fixes many, many many things!!!

perfomance is really enhanced (test it with glxgears for example)

Many issues with video like glxgears, googleearth, compiz, and things like that are fixed!!

I suggested this version of driver should be default driver for Jaunty and others but somebody rejected my opinion and said to me "if you want to get latest driver install it from x-updates"

Then I'm trying to resolve this myself.

If you want to enable compiz in Jaunty you can do it adding this SKIP_CHECKS=yes
 to one of these files:

~/.config/compiz/compizconfig/config
~/.config/compiz/compiz-manager

I added the line to both files because I'm not sure wich one is correct in Ubuntu, but at least it is working really great than before!

[Pablo Estigarribia (pablodav)]

Look this screen is working with xserver-xorg-video-intel 2.7 and UXA, It's amazing! It never was possible for me before!

[Robert Persson (ireneshusband)]

Synaptic complains that it can't find the key for the x-updates repository. I have in fact successfully installed the correct key 0x3b22ab97af1cdfa9. The key that Synaptic is complaining about however is 0x5EB2EDFE1FCB21DB, i.e. a completely different key. Can anyone tell me what to do here?

[Julien Olivier (julo)]

I've just tested with XUA enable, DRI2 enabled and UseFBDev enabled and with the packages coming from x-updates on this device: Intel Corporation 82852/855GM Integrated Graphics Device [8086:3582] (rev 02). The result is really disappointing:
 - the bug with videos when compiz is running is still there
 - glxgears no run at about 300 fps instead of 500 fps before (without compiz enabled)
 - if I enable compiz, glxgears drop to 200 fps

PS: I submitted an entry to http://wiki.ubuntu.com/X/UxaTesting

[Launchpad Janitor (janitor)]

This bug was fixed in the package xserver-xorg-video-intel - 2:2.7.99.1+git20090602.ec2fde7c-0ubuntu1

---------------
xserver-xorg-video-intel (2:2.7.99.1+git20090602.ec2fde7c-0ubuntu1) karmic; urgency=low

  * Update to git 20090602 (master branch) up to commit ec2fde7c
    - xvmc is disabled since DRI1 no longer supported
    - LP: #96991 - 3D stuff breaks with Compiz: Redirected Direct Rendering
      is needed in DRI
    - LP: #120834 - X freezes with I830WaitLpRing error when running OpenGL apps
    - LP: #337608 - X crashes in fbBlt() when using Sun Java Plugin 6 + firefox3.0
    - LP: #339555 - compiz slowmotion after Jaunty upgrade
    - LP: #363900 - X.org freezes with intel driver, no apparent trigger
    - LP: #331719 - VT switching doesn't work on Intel 915GM
    - LP: #339091 - X freezes a few minutes after resuming
    - LP: #348436 - Kubuntu: X server crash when screensaver is started (4500MHD)
    - LP: #279727 - Kubuntu: Display Corruption w/ Intel 4700MHD
    - LP: #357851 - Kubuntu: Distorted display after switching virtual desktops w/ exa
    - LP: #158415 - Front buffer dynamic resize not supported
    - LP: #324998 - x server restarts itself w/ compiz on Intel 945GM
    - LP: #355593 - after upgrade to 9.04, rotating desktop cube ran slow
    - LP: #357290 - 1 fps in 3d apps like neverball with EXA
    - LP: #360774 - Graphical Corruption with EXA on X4500
    - LP: #364126 - screensaver prefs dialog in 9.04 RC livecd leaves dirt
    - LP: #375712 - Native resolution for dell "2005fpw" monitor not listed
    - LP: #375264 - Choppy flash video and poor performance with compiz
    - LP: #349568 - Jaunty / Compiz slow and tearing on GMA 4500MHD
    - LP: #356056 - window tearing during movement on 965 (no compiz)
    - LP: #330460 - xorg shows black image/hangs with jpg in firefox
    - LP: #347587 - X asserts on pI830->batch_ptr != 0 on resume from suspend
  * Merge with Debian experimental. Remaining Ubuntu changes:
    - Add lpia architecture
    - Re-enable the patch system, add quilt to build-deps.
    - 110_quirk_hp_mini.patch: quirk (sent upstream)
    - 117_quirk_thinkpad_x30.patch: quirk (sent upstream)
  * Drop 116_8xx_disable_dri.patch. There have been fixes for 3d on 8xx
    chipsets upstream, so drop the DRI disablement so the fixes can be
    re-tested.
  * Drop 103_quirk_intel_mb890.patch. Better quirk available upstream.
    (LP: #305269)

 -- Bryce Harrington <email address hidden> Tue, 02 Jun 2009 10:47:32 -0700

[Derek White (d-man97)]

Regarding ati/radeon and the video not moving with the window when compiz is used, I have a workaround:

[I am using medibuntu's mplayer, I assume it will work with the normal one too. My xorg.conf is blank.]
Open video in mplayer, let it play, & stop the video - leave mplayer open. Open video in a new instance of mplayer: the video stays in its window when moved, all compiz effects (Alt-Tab, wobble, cube, etc.) work with the video staying inside its window. The original mplayer instance can be closed at any time after the second has started. [This might work in totem as well, but I do not know how to open multiple instances of totem.]

Once all instances of mplayer are closed, you have to re-do the workaround for it to work again.

[Derek White (d-man97)]

I also agree with: Oliver Mattos' postings from 2008-04-27 and 2008-07-18; Vytas' post from 2008-10-03 "...enabling obviously crippled software does more harm than good..." - just look at the dups; & David Fokkema's post on 2008-10-05. Compiz should not be enabled by default with ati driver until this is fixed.

Having a default not work properly when simply disabling the Visual Effects setting makes it work is annoying to users. Disabling it by default could prevent the user from being disappointed/annoyed. Then, when they activate it, it's their own fault for messing with things. Isn't that the Linux way? ;)

[Extender (msveshnikov)]

So, for ATI X700, UXA/DRI2 still not working?
I have this in Xorg.log:
 AIGLX: Screen 0 is not DRI2 capable

xorg.conf is configured correctly, I tried kernels 2.6.30 and 2.6.31rc3 and radeon driver from Xorg-edgers PPA.

[Vish (vish)]

Architecture: i386
DistroRelease: Ubuntu 9.10
Uname: Linux 2.6.31-4-generic i686
ATI Mobility Raedon X1400

[Vish (vish)]

Still present in :
Architecture: i386
DistroRelease: Ubuntu 9.10
Uname: Linux 2.6.31-4-generic i686
ATI Mobility Raedon X1400

[Launchpad Janitor (janitor)]

This bug was fixed in the package xserver-xorg-video-ati - 1:6.12.99+git20090825.fc74e119-0ubuntu1

---------------
xserver-xorg-video-ati (1:6.12.99+git20090825.fc74e119-0ubuntu1) karmic; urgency=low

  * Checkout from git 20090825 (master branch) up to commit
    fc74e1194c980d978667e02c60a29a761a694bde
    + Adds DRI2 / redirected direct rendering
      (LP: #96991)
    + Fix freeze on opengl games (bzflag, alienarena) (when KMS on)
      (LP: #348450)
    + Fix intermittent short screen blanking behavior
      (LP: #310864)
    + Fix faulty wine screen updates with EXA Compositing
      (LP: #314205)
    + Fix screen corruption issues when switching windows
      (LP: #406731)
    + Fix SIGSEGV crash in drmCommandNone()
      (LP: #352567)
    + Fix smeared/tearing of display on reboot
      (LP: #367741

  [Tormod Volden]
  * 199_add_git_version_to_log.diff: Log git commit id in RadeonPreInit()
  * sed -i s/DRI2BufferPtr/DRI2Buffer2Ptr/ src/radeon_dri2.c
      so it compiles against xserver 1.6.2.+

  [Bryce Harrington]
  * Restore patch system and quilt to build-depends
  * Drop patches already present upstream
    + 107_check_unsupported_composit_ops.patch
    + 108_quirk_agpmode_m6_ali.patch: AGPMode quirk.
    + 109_quirk_agpmode_m7_intel.patch: AGPMode quirk.
    + 110_quirk_agpmode_r420_sis.patch: AGPMode quirk.
    + 111_use_xaa_for_lowmem_or_nodri.patch

 -- Bryce Harrington <email address hidden> Sat, 29 Aug 2009 12:16:53 -0700

[Extender (msveshnikov)]

And how to test that?

[Tormod Volden (tormodvolden)]

You can test it in Karmic (or the Karmic alpha 5 live CD, due tomorrow) by adding the boot parameter "radeon.modeset=1".

[Derek White (d-man97)]

Using the Alpha 5 Live CD, I pressed F6 to edit the options and added
radeon.modeset=1 just before the 'quiet splash -- ' parameters. Once the
boot process started, the screen said something like "unknown boot
option, ignoring" in reference to the modeset option. Boot process
continued normally, until after the boot splash went away (prior to gdm
starting): my screen went into power saving mode and the computer did
nothing. Ctrl-Alt-Del had no effect and I had to hard reset the system.

I am using an ATI Radeon 9550 (RV350) 256MB graphics card (manufactured
by ATI) on a Biostar M7NCD motherboard with an nVidia nForce2 400 + MCP
chipset. If there is any further testing/debugging I can do from the
Live CD, please let me know.

[Vish (vish)]

"radeon.modeset=1" needs to be added at the end of the kernel line, 1 space after "splash". [Well, that's how it works for me :) ]
But do note , there is regression with using KMS , the FPS will be much lower than without the KMS. [in my case nearly half]
So the system might *seem* a bit sluggish, well it is actually the slow rendering and not the cpu slowness.

For any other info > https://wiki.ubuntu.com/X or you can ask for help in irc #ubuntu-x .

[Tormod Volden (tormodvolden)]

You can put radeon.modeset=1 anywhere on the line. But see bug #410058 instead for this ati-specific issue.

[Eugene Savelov (savelov)]

Tested with Radeon XPRESS 200M - when in DRI2 mode, with KMS, Google earth have visual problems
without KMS, displays correctly

[Reynaldo Matos Hortensi (lasanhadeberinjela)]

I have this bug on the oficial release of Karmic Koala using amd64 desktop edition on a radeon xpress 200 (RS482).
Some help is needed.
Thanks

[Reynaldo Matos Hortensi (lasanhadeberinjela)]

If I run:
strace glxgears

the system (or xorg) freezes completely, but if I run:
strace -o glxgears.trace.log glxgears

it doesn't.
Maybe Its related to this bug.
May I open another bug?

[Reynaldo Matos Hortensi (lasanhadeberinjela)]

the "strace glxgears issue" happens too using the proprietary driver (fglrx), but instead of karmic in intrepid.

[Alexander Jones (alex-weej)]

I can assure you this has nothing to do with this bug. Please discuss
this elsewhere!

2009/11/10 Reynaldo Matos Hortensi <email address hidden>:
> the "strace glxgears issue" happens too using the proprietary driver
> (fglrx), but instead of karmic in intrepid.
>
> --
> (Needs radeon-rewrite) 3D stuff breaks with Compiz:  Redirected Direct Rendering is needed in DRI
> https://bugs.launchpad.net/bugs/96991
> You received this bug notification because you are a direct subscriber
> of the bug.
>

[Extender (msveshnikov)]

Yes I can confirm that bug is fully resolved (ATI X700) with radeon.modeset=1 kernel parameter.
Windows overlap normal with Compiz on and off.
Gogle earth has NO artefacts and flickering (with Compiz on!)
Performance of glxgears is slightly decreased (1800 fps against 2400 fps).

[Reynaldo Matos Hortensi (lasanhadeberinjela)]

Hi Extender,
I have tried with xorg-edger drivers and radeon.modset=1 kernel parameter... everything worked fine but googleearth (ATI Radeon Xpress 200, chipset RS482).
Glxgears worked fine (I can move the glxgears window without artefacts on screen).
Googleearth have produced artefacts (see attachment).
I will be back with number for glxgears and gtkperf.

[Pauli (paniemin)]

This is different problem. This looks a lot like rendering bug that should
be reported in new bug report (unless it is already reported in some newer
bug report)

On Fri, Dec 25, 2009 at 1:56 PM, Reynaldo Matos Hortensi <
<email address hidden>> wrote:

> Hi Extender,
> I have tried with xorg-edger drivers and radeon.modset=1 kernel
> parameter... everything worked fine but googleearth (ATI Radeon Xpress 200,
> chipset RS482).
> Glxgears worked fine (I can move the glxgears window without artefacts on
> screen).
> Googleearth have produced artefacts (see attachment).
> I will be back with number for glxgears and gtkperf.
>
>

[Freedesktop-org (freedesktop-org)]

*** Bug 27947 has been marked as a duplicate of this bug. ***

[Freedesktop-org (freedesktop-org)]

Why is this marked as FIXED? I see that one poster found a workaround by manually enabling dri2 and uxa on Ubuntu, but the fix was never ported to the released version of the driver. That is, even on Ubuntu, users are bitten by this bug unless the know how to enable dri2 and uxa, and assuming that they know that they even have to.

I am reopening until the issue is fixed, a workaround (or manually enabling dri2/uxa) is not a fix.

[agd5f (agd5f)]

One of the main points of DRI2 was to fix GL + composite. It will only work as expected with a DRI2 capable driver. DRI2 is not a workaround, it is the fix.

[Freedesktop-org (freedesktop-org)]

> One of the main points of DRI2 was to fix GL + composite. It
> will only work as expected with a DRI2 capable driver. DRI2 is
> not a workaround, it is the fix.

I have misunderstood, then. I understood this issue to be for improving OpenGL (and DRI) support for a specific driver. I was too quick to mark my own bug as a dupe (see comment 25). I will reopen that bug as a request to enable DRI in radeonhd.

Thanks.

[Julien Cristau (jcristau)]

> --- Comment #28 from Dotan Cohen <email address hidden> 2010-05-03 12:19:42 PDT ---
> I will reopen that bug as a request to enable DRI in radeonhd.
>
don't. just use radeon instead.

[suoko (suoko)]

LinkedIn
------------

Bug,

Vorrei aggiungerti alla mia rete professionale su LinkedIn.

-gabriele

gabriele vidali
IT helpdesk presso Temera Srl
Firenze, Italia

Conferma che conosci gabriele vidali:
https://www.linkedin.com/e/-7gpsk-ha2yz419-3d/isd/9808074698/qAAR2ok5/?hs=false&tok=31Z9G8hHMbo5w1

--
Stai ricevendo inviti di collegamento. Clicca per annullare l'iscrizione:
http://www.linkedin.com/e/-7gpsk-ha2yz419-3d/Xk1OvfwjxCYck4FyPC5cBSoye1ERMckp7JxCqy/goo/96991%40bugs%2Elaunchpad%2Enet/20061/I3274259603_1/?hs=false&tok=2QRnqoBGIbo5w1

(c) 2012 LinkedIn Corporation. 2029 Stierlin Ct, Mountain View, CA 94043, USA.