[SOLVED]Xf86-video-intel 2.7.0-1 problem.

lycaeum · 2009-04-17 00:33:02

Iḿ using xf86-video-intel with xorg-server 1.6.1 and when i use a kms-enabled kernel my video gets all messed up. It' s like it can't find the correct sync with video.

no kms is ok, but as usual , i get no direct rendering with compiz.

does anyone is experiencing the same problem?

intel gm965 here

Last edited by lycaeum (2009-04-17 10:50:22)

GogglesGuy · 2009-04-17 01:12:22

lycaeum wrote:

Iḿ using xf86-video-intel with xorg-server 1.6.1 and when i use a kms-enabled kernel my video gets all messed up. It' s like it can't find the correct sync with video.
no kms is ok, but as usual , i get no direct rendering with compiz.
does anyone is experiencing the same problem?
intel gm965 here

Not sure if it's the same problem, but I had to disable tiling in order to get it to display proper on my gm965.

lycaeum · 2009-04-17 01:27:20

And how do i disable tiling? never heard of it.

GogglesGuy · 2009-04-17 02:16:53

lycaeum wrote:

And how do i disable tiling? never heard of it.

In your /etc/X11/xorg.conf (create one if you don't have one), add a Option "Tiling" "false"

Section "Device"
    Identifier "mydevice"
    Driver "intel"
    Option "Tiling" "false"
EndSection

Also check the intel man page:

man intel

dakor · 2009-04-17 02:22:54

lycaeum wrote:

Iḿ using xf86-video-intel with xorg-server 1.6.1 and when i use a kms-enabled kernel my video gets all messed up. It' s like it can't find the correct sync with video.
no kms is ok, but as usual , i get no direct rendering with compiz.
does anyone is experiencing the same problem?
intel gm965 here

same prob

dakor · 2009-04-17 02:30:17

Quick question: With this new driver and with KMS and UXA enabled is the "Option "MigrationHeuristic" "greedy"" option still needed?

GogglesGuy · 2009-04-17 03:03:34

dakor wrote:

Quick question: With this new driver and with KMS and UXA enabled is the "Option "MigrationHeuristic" "greedy"" option still needed?

That seems to be a EXA option (man exa), so I'm not sure if UXA will use that...

xodeus · 2009-04-17 10:09:09

your'e talking about creating a xorg.conf file. Should it only contain this?

Section "Device"
    Identifier "mydevice"
    Driver "intel"
    Option "Tiling" "false"
EndSection

lycaeum · 2009-04-17 10:49:40

GogglesGuy wrote:

lycaeum wrote:
And how do i disable tiling? never heard of it.
In your /etc/X11/xorg.conf (create one if you don't have one), add a Option "Tiling" "false"
Section "Device"
    Identifier "mydevice"
    Driver "intel"
    Option "Tiling" "false"
EndSection
Also check the intel man page:
man intel

Thank you! It worked like a charm

GogglesGuy · 2009-04-17 13:07:07

xodeus wrote:

your'e talking about creating a xorg.conf file. Should it only contain this?
Section "Device"
    Identifier "mydevice"
    Driver "intel"
    Option "Tiling" "false"
EndSection

That's what I did. The rest should be automatically set by Xorg/HAL.

Last edited by GogglesGuy (2009-04-18 01:47:48)

dgr48 · 2009-04-17 13:51:40

And also the poor performance in compiz!!!!

Benchmark indicates with the legacy drivers: 96 fps
With the 'updated' one: 16 fps

Something went wrong with the new release.

Card: Intel Corporation Mobile 945GM/GMS, 943/940GML Express Integrated Graphics Controller

Regards.

Bogart · 2009-04-17 16:29:15

The problem is that tiling was enabled to solve the performance problems with KMS. So disabling it will have the performance problems again.

How does gtkperf run as compared with EXA when using KMS and tiling off? Could anyone test? (I'll test myself later when I have the chance).

GogglesGuy · 2009-04-17 17:11:21

Bogart wrote:

The problem is that tiling was enabled to solve the performance problems with KMS. So disabling it will have the performance problems again.
How does gtkperf run as compared with EXA when using KMS and tiling off? Could anyone test? (I'll test myself later when I have the chance).

Don't know how it compares, but here are the UXA/kms results with tiling off:

GtkPerf 0.40 - Starting testing: Fri Apr 17 12:10:03 2009

GtkEntry - time:  0.00
GtkComboBox - time:  3.09
GtkComboBoxEntry - time:  1.87
GtkSpinButton - time:  0.89
GtkProgressBar - time:  1.03
GtkToggleButton - time:  0.61
GtkCheckButton - time:  0.53
GtkRadioButton - time:  0.71
GtkTextView - Add text - time:  0.91
GtkTextView - Scroll - time:  0.88
GtkDrawingArea - Lines - time:  1.23
GtkDrawingArea - Circles - time:  2.42
GtkDrawingArea - Text - time:  2.37
GtkDrawingArea - Pixbufs - time:  0.27
 --- 
Total time: 16.84

Denacke · 2009-04-17 19:27:02

For me, tiling also doesn't work with the latest kernel, but it does with one release back (-2 instead of -3).

You could check http://bbs.archlinux.org/viewtopic.php?id=69695 for more information about that.

I uploaded the previous version of the kernel in case anyone doesn't have it and would like to test it: http://denacke.net/archlinux/kernel3/ke … pkg.tar.gz

GogglesGuy · 2009-04-17 19:36:36

Denacke wrote:

For me, tiling also doesn't work with the latest kernel, but it does with one release back (-2 instead of -3).
You could check http://bbs.archlinux.org/viewtopic.php?id=69695 for more information about that.
I uploaded the previous version of the kernel in case anyone doesn't have it and would like to test it: http://denacke.net/archlinux/kernel3/ke … pkg.tar.gz

I'm not sure if that is totally related. My driver reports success when allocating a tiled buffer.

(II) intel(0): Tiled allocation successful.

So I think it is rather a bug in the driver using the tiled buffer.

Denacke · 2009-04-17 19:44:48

I also get that, but in that same log it'll also say:

(EE) intel(0): Failed to set tiling on front buffer: rejected by kernel
(EE) intel(0): Failed to set tiling on back buffer: rejected by kernel
(EE) intel(0): Failed to set tiling on depth buffer: rejected by kernel

I don't get those lines in that previous kernel version. Mind checking if you also get those?

GogglesGuy · 2009-04-17 20:07:53

Denacke wrote:

I also get that, but in that same log it'll also say:
(EE) intel(0): Failed to set tiling on front buffer: rejected by kernel
(EE) intel(0): Failed to set tiling on back buffer: rejected by kernel
(EE) intel(0): Failed to set tiling on depth buffer: rejected by kernel
I don't get those lines in that previous kernel version. Mind checking if you also get those?

Nope, don't have those.

Bogart · 2009-04-17 20:32:46

GogglesGuy wrote:

Don't know how it compares, but here are the UXA/kms results with tiling off:
GtkPerf 0.40 - Starting testing: Fri Apr 17 12:10:03 2009

 --- 
Total time: 16.84

That means the performance problem is fixed even without tiling. I tested myself too and got 8.5 secs (on a X4500HD intel card), while with the previous version of the driver it could take minutes if using KMS.

Denacke · 2009-04-17 20:55:07

I get

Total time:  8.62

with tiling enabled (and that previous kernel version of course) and a pretty outdated Intel Corporation Mobile 915GM/GMS/910GML card and no KMS.

GogglesGuy · 2009-04-17 21:04:35

Bogart wrote:

GogglesGuy wrote:
Don't know how it compares, but here are the UXA/kms results with tiling off:
GtkPerf 0.40 - Starting testing: Fri Apr 17 12:10:03 2009

 --- 
Total time: 16.84
That means the performance problem is fixed even without tiling. I tested myself too and got 8.5 secs (on a X4500HD intel card), while with the previous version of the driver it could take minutes if using KMS.

I also tested it with the tiling on, but I'm not sure if the numbers mean anything since it's broken:

Total time:  5.09

Bogart · 2009-04-17 21:44:59

GogglesGuy wrote:

I also tested it with the tiling on, but I'm not sure if the numbers mean anything since it's broken:
Total time:  5.09

I don't know either, but it seems that today a fix for the tiling problem went into the kernel, so hopefully it will be backported to the next stable release.

edubarr · 2009-04-17 23:19:47

Thanks for the tips. I was having a lot of problems with the xf86-video-intel 2.6 package lagging a lot with xfig and inkscape. I tried with the drivers in testing with tiling disabled and everything works fine.

mwc · 2009-04-20 06:37:01

This bug is apparently fixed, tiling in 2.7.0 requires a kernel patch.

http://bugs.freedesktop.org/show_bug.cgi?id=21246#c3

Matt.

Heller_Barde · 2009-04-20 09:21:15

I'm putting together this patch to try out, if it works, i'll post the pkgbuild here

Heller_Barde · 2009-04-21 12:15:42

if anyone wants to know, it does not just work with the patch integrated into core kernel26
but it was fun to try. at least i know now how to make a patch
cheers
Barde

Arch Linux

#1 2009-04-17 00:33:02

[SOLVED]Xf86-video-intel 2.7.0-1 problem.

#2 2009-04-17 01:12:22

Re: [SOLVED]Xf86-video-intel 2.7.0-1 problem.

#3 2009-04-17 01:27:20

Re: [SOLVED]Xf86-video-intel 2.7.0-1 problem.

#4 2009-04-17 02:16:53

Re: [SOLVED]Xf86-video-intel 2.7.0-1 problem.

#5 2009-04-17 02:22:54

Re: [SOLVED]Xf86-video-intel 2.7.0-1 problem.

#6 2009-04-17 02:30:17

Re: [SOLVED]Xf86-video-intel 2.7.0-1 problem.

#7 2009-04-17 03:03:34

Re: [SOLVED]Xf86-video-intel 2.7.0-1 problem.

#8 2009-04-17 10:09:09

Re: [SOLVED]Xf86-video-intel 2.7.0-1 problem.

#9 2009-04-17 10:49:40

Re: [SOLVED]Xf86-video-intel 2.7.0-1 problem.

#10 2009-04-17 13:07:07

Re: [SOLVED]Xf86-video-intel 2.7.0-1 problem.

#11 2009-04-17 13:51:40

Re: [SOLVED]Xf86-video-intel 2.7.0-1 problem.

#12 2009-04-17 16:29:15

Re: [SOLVED]Xf86-video-intel 2.7.0-1 problem.

#13 2009-04-17 17:11:21

Re: [SOLVED]Xf86-video-intel 2.7.0-1 problem.

#14 2009-04-17 19:27:02

Re: [SOLVED]Xf86-video-intel 2.7.0-1 problem.

#15 2009-04-17 19:36:36

Re: [SOLVED]Xf86-video-intel 2.7.0-1 problem.

#16 2009-04-17 19:44:48

Re: [SOLVED]Xf86-video-intel 2.7.0-1 problem.

#17 2009-04-17 20:07:53

Re: [SOLVED]Xf86-video-intel 2.7.0-1 problem.

#18 2009-04-17 20:32:46

Re: [SOLVED]Xf86-video-intel 2.7.0-1 problem.

#19 2009-04-17 20:55:07

Re: [SOLVED]Xf86-video-intel 2.7.0-1 problem.

#20 2009-04-17 21:04:35

Re: [SOLVED]Xf86-video-intel 2.7.0-1 problem.

#21 2009-04-17 21:44:59

Re: [SOLVED]Xf86-video-intel 2.7.0-1 problem.

#22 2009-04-17 23:19:47

Re: [SOLVED]Xf86-video-intel 2.7.0-1 problem.

#23 2009-04-20 06:37:01

Re: [SOLVED]Xf86-video-intel 2.7.0-1 problem.

#24 2009-04-20 09:21:15

Re: [SOLVED]Xf86-video-intel 2.7.0-1 problem.

#25 2009-04-21 12:15:42

Re: [SOLVED]Xf86-video-intel 2.7.0-1 problem.

Board footer