Testing latest -intel in Ubuntu

[kxmas]

I should also mention to anyone interested that there are unofficial Ubuntu packages for the 2.6.29 kernels at https://wiki.ubuntu.com/KernelMainlineBuilds . I would recommend using those when trying out the latest -intel drivers.

I'm missing something here. I'm running the 2.6.29-020629rc7-generic kernel from the MainlineBuilds and I'm using drm-snapshot and intel drivers from the xorg-edgers ppa. Before I used module-assistant to build the drm-modules, I saw a message about about mrtt not being able to allocate something and it might hinder performance. After I used module-assistant to build drm-modules, the i915 kernel module refuses to load.

I have the kernel source, but the makefile is rather empty. What's the quickest way to just rebuild the i915 module without building everything?

[tormod]

I'm missing something here. I'm running the 2.6.29-020629rc7-generic kernel from the MainlineBuilds and I'm using drm-snapshot and intel drivers from the xorg-edgers ppa. Before I used module-assistant to build the drm-modules, I saw a message about about mrtt not being able to allocate something and it might hinder performance. After I used module-assistant to build drm-modules, the i915 kernel module refuses to load.

I have the kernel source, but the makefile is rather empty. What's the quickest way to just rebuild the i915 module without building everything?

If you are on intel, you should not use the drm-module-source package, since the i915 module is not maintained in linux-core (the kernel module tree in libdrm, which is packaged in drm-module-source) any longer. The 2.6.29 stock modules will be more up to date.

If you want on-the-edge kernel modules for intel (beyond stock 2.6.29), you will have to get the drm-intel-next branch of the kernel sources, see http://thread.gmane.org/gmane.comp.v...ri.devel/33449 . There are ways to compile just parts of the kernel source (like only the drm modules), see http://thread.gmane.org/gmane.comp.v...75/focus=33273

[kxmas]

If you are on intel, you should not use the drm-module-source package, since the i915 module is not maintained in linux-core (the kernel module tree in libdrm, which is packaged in drm-module-source) any longer. The 2.6.29 stock modules will be more up to date.

If you want on-the-edge kernel modules for intel (beyond stock 2.6.29), you will have to get the drm-intel-next branch of the kernel sources, see http://thread.gmane.org/gmane.comp.v...ri.devel/33449 . There are ways to compile just parts of the kernel source (like only the drm modules), see http://thread.gmane.org/gmane.comp.v...75/focus=33273

Thanks for the info. i915 from drm-intel-next wouldn't compile against the rc7 source. I'm just going to leave as it is.

[Ant P.]

I'm playing with driver version 2.6.1 on my eee701 (Debian, experimental), and for the most part it's been okay, modesetting works great too. I've got a few things really annoying me though:

1. It defaults to a 1360x1360 virtual screen (complete with mouse disappearing offscreen) unless I force "Virtual 800 480".
2. It seems to be using software OpenGL even though the DRI stuff is all installed and Xorg.0.log shows no problems.
3. It's running in 6-bit mode. Ew. There's visible banding on images that the 2.3 driver didn't have.
(and #4: upgrading this required a new synaptics driver that screwed up everything, but let's not go there...)

Are there some settings I need to change to fix these?

[mrvanes]

xserver-xorg-video-intel 2:2.6.99.1+git20090308.73aa23d9-0ubuntu0tormod seems broken. I had to revert back to Jaunty's 2:2.6.1-1ubuntu4 to make X work again.

Both kernel 2.6.28.7 and 2.6.29-rc7 trigger this problem.

Tail of the failing Xorg.0.log.old
[ 0.212468] (II) intel(0): Modeline "720x400"x85.0 35.50 720 756 828 936 400 401 404 446 -hsync +vsync (37.9 kHz)
[ 0.212478] (II) intel(0): Modeline "640x400"x85.1 31.50 640 672 736 832 400 401 404 445 -hsync +vsync (37.9 kHz)
[ 0.212488] (II) intel(0): Modeline "640x350"x85.1 31.50 640 672 736 832 350 382 385 445 +hsync -vsync (37.9 kHz)

Backtrace:
0: /usr/bin/X(xorg_backtrace+0x26) [0x4f1916]
1: /usr/bin/X(xf86SigHandler+0x41) [0x485a91]
2: /lib/libc.so.6 [0x7fbc5ae10040]
3: /usr/bin/X(dixAllocatePrivate+0x79) [0x434689]
4: /usr/bin/X(dixLookupPrivate+0x35) [0x434875]
5: /usr/lib/xorg/modules/extensions//libdri.so(DRIGetSAREAPrivate+0x17) [0x7fbc596e5677]
6: /usr/lib/xorg/modules/drivers//intel_drv.so(i830_update_dri_buffers+0x39) [0x7fbc5908b199]
7: /usr/bin/X(xf86CrtcSetModeTransform+0x47d) [0x4ac61d]
8: /usr/lib/xorg/modules/drivers//intel_drv.so(i830GetLoadDetectPipe+0x13c) [0x7fbc59052d1c]
9: /usr/lib/xorg/modules/drivers//intel_drv.so [0x7fbc5906dd80]
10: /usr/bin/X(xf86ProbeOutputModes+0x1a5) [0x4ad015]
11: /usr/bin/X(xf86InitialConfiguration+0x17d) [0x4adaed]
12: /usr/lib/xorg/modules/drivers//intel_drv.so [0x7fbc590562c0]
13: /usr/lib/xorg/modules/drivers//intel_drv.so [0x7fbc5905a667]
14: /usr/bin/X(InitOutput+0xdbc) [0x46f9cc]
15: /usr/bin/X(main+0x20e) [0x433c2e]
16: /lib/libc.so.6(__libc_start_main+0xe6) [0x7fbc5adfb5a6]
17: /usr/bin/X [0x433269]
Saw signal 11. Server aborting.
ddxSigGiveUp: Closing log

[kxmas]

xserver-xorg-video-intel 2:2.6.99.1+git20090308.73aa23d9-0ubuntu0tormod seems broken. I had to revert back to Jaunty's 2:2.6.1-1ubuntu4 to make X work again.

Both kernel 2.6.28.7 and 2.6.29-rc7 trigger this problem.

Tail of the failing Xorg.0.log.old
[ 0.212468] (II) intel(0): Modeline "720x400"x85.0 35.50 720 756 828 936 400 401 404 446 -hsync +vsync (37.9 kHz)
[ 0.212478] (II) intel(0): Modeline "640x400"x85.1 31.50 640 672 736 832 400 401 404 445 -hsync +vsync (37.9 kHz)
[ 0.212488] (II) intel(0): Modeline "640x350"x85.1 31.50 640 672 736 832 350 382 385 445 +hsync -vsync (37.9 kHz)

Backtrace:
0: /usr/bin/X(xorg_backtrace+0x26) [0x4f1916]
1: /usr/bin/X(xf86SigHandler+0x41) [0x485a91]
2: /lib/libc.so.6 [0x7fbc5ae10040]
3: /usr/bin/X(dixAllocatePrivate+0x79) [0x434689]
4: /usr/bin/X(dixLookupPrivate+0x35) [0x434875]
5: /usr/lib/xorg/modules/extensions//libdri.so(DRIGetSAREAPrivate+0x17) [0x7fbc596e5677]
6: /usr/lib/xorg/modules/drivers//intel_drv.so(i830_update_dri_buffers+0x39) [0x7fbc5908b199]
7: /usr/bin/X(xf86CrtcSetModeTransform+0x47d) [0x4ac61d]
8: /usr/lib/xorg/modules/drivers//intel_drv.so(i830GetLoadDetectPipe+0x13c) [0x7fbc59052d1c]
9: /usr/lib/xorg/modules/drivers//intel_drv.so [0x7fbc5906dd80]
10: /usr/bin/X(xf86ProbeOutputModes+0x1a5) [0x4ad015]
11: /usr/bin/X(xf86InitialConfiguration+0x17d) [0x4adaed]
12: /usr/lib/xorg/modules/drivers//intel_drv.so [0x7fbc590562c0]
13: /usr/lib/xorg/modules/drivers//intel_drv.so [0x7fbc5905a667]
14: /usr/bin/X(InitOutput+0xdbc) [0x46f9cc]
15: /usr/bin/X(main+0x20e) [0x433c2e]
16: /lib/libc.so.6(__libc_start_main+0xe6) [0x7fbc5adfb5a6]
17: /usr/bin/X [0x433269]
Saw signal 11. Server aborting.
ddxSigGiveUp: Closing log

Yep, xserver-xorg-video-intel 2:2.6.99.1+git20090308.73aa23d9-0ubuntu0tormod is no good. I rolled back to 2:2.6.99.1+git20090306.67fef27f-0ubuntu0tormod. I reverted all my changes until rolling back the video driver was the only thing left.

I'm using Intel Corporation Mobile GM965/GL960 Integrated Graphics Controller (rev 0c).

[tormod]

xserver-xorg-video-intel 2:2.6.99.1+git20090308.73aa23d9-0ubuntu0tormod seems broken. I had to revert back to Jaunty's 2:2.6.1-1ubuntu4 to make X work again.

Anyone else having tested these packages?

mrvanes, can you please install -dbg packages so that you get a full symbolic backtrace?

Did you try the previous snapshot? The difference includes a recent commit
that involves DRIGetSAREAPrivate, which shows up in the backtrace.

Update: Just saw kxmas' post. Thanks. The reason is probably http://cgit.freedesktop.org/xorg/dri...b910acd164abf5
But are you not using DRI2 instead of DRI1?

[mrvanes]

Although it's against all advises, I'm running Jaunty and your intel drivers on a production laptop, so I'm a bit reluctant to knowingly break my system (on a part that's quite important for everyday use
And FYI: I'm using DRI2 (UXA) as a default in my xorg.conf. Switching to EXA didn't help with the broken driver.

[tormod]

I'm playing with driver version 2.6.1 on my eee701 (Debian, experimental), and for the most part it's been okay, modesetting works great too. I've got a few things really annoying me though:

1. It defaults to a 1360x1360 virtual screen (complete with mouse disappearing offscreen) unless I force "Virtual 800 480".
2. It seems to be using software OpenGL even though the DRI stuff is all installed and Xorg.0.log shows no problems.
3. It's running in 6-bit mode. Ew. There's visible banding on images that the 2.3 driver didn't have.
(and #4: upgrading this required a new synaptics driver that screwed up everything, but let's not go there...)

Are there some settings I need to change to fix these?

For 3) Some panels are really 6 bit, http://lists.freedesktop.org/archive...ch/044192.html

For kernel modesetting, which kernel are you using? And libdrm?

BTW, For Debian, you can also download the xorg-edgers source package of -intel and rebuild it on your system, if you want to try out the trunk, but you will need libdrm 2.4.5.

[tormod]

Although it's against all advises, I'm running Jaunty and your intel drivers on a production laptop,

That's the spirit I think I have been guilty in doing similar things.

so I'm a bit reluctant to knowingly break my system (on a part that's quite important for everyday use
And FYI: I'm using DRI2 (UXA) as a default in my xorg.conf. Switching to EXA didn't help with the broken driver.

Maybe kxmas can help with the symbolic backtrace? Anyway, if 20090306 works but 20090308 fails this looks like an upstream regression that should be reported. They will likely ask for full logs, and a symbolic backtrace if possible.

Well, I am looking at the code, and if you are running DRI2, i830_update_dri_buffers() should not call DRIGetSAREAPrivate() like it does in your backtrace.

See also https://wiki.ubuntu.com/X/Backtracing for how to get a full backtrace.