You are not logged in.

#1 2008-09-23 12:14:02

jwcxz
Member
Registered: 2008-09-23
Posts: 239
Website

xf86-video-intel 2.4.2-1 has "spontaneous" lockups

I've been quite happy with my Arch installation on my Dell XPS M1330, which I purchased about a month ago.  I have used Arch quite a bit before for servers and virtual machines, but this is my first time using it as a "regular" desktop system.

Anyways, I have had lots of problems with the integrated Intel GM965 graphics chipset.  It seems as though no matter what driver version I'm using, EXA has lots of graphical glitches (e.g. the system tray in KDE 4 is not viewable because there are seemingly random pieces of the screen splotched over it) and 3D support under both EXA and XAA seems to be minimal ("Failed to initialize TTM buffer manager.  Falling back to classic." on everything, low FPS in glxgears, absurdly poor performance on 3D games, etc.).  Right now, I'm using XAA and RandR to support my dual-screen setup (the included laptop monitor runs at 1280x800 and my external monitor runs at 1280x1024).

But none of that seems important right now compared to the fact that the xf86-video-intel driver is causing seemingly arbitrary lockups.  Most of the time, after the X server crashes, my whole system proceeds to crash as well and I need to manually shut off the power, which makes me very worried about data corruption.

I put spontaneous in quotes in the subject because I have a theory as to why this is happening.  The KDM log tells me this:
http://pastebin.com/f2019f021
(that's one X session; don't mind the stuff about the input devices--I haven't quite set those up yet)

I searched for errors with I830WaitLpRing() and came up with a very large number of results, which suggests that it could be just a blanket problem.  One interesting note was that someone had mentioned lockups occurring while using Xfig.  Interestingly enough, I was using Xfig when my system last locked up as well.  However, I've also had lockups when running virtual machines (VirtualBox and/or VMWare) and sometimes when just browsing sites with a lot of Flash content.  It's always this same error, with some minor variations with respect to the "wanted line":

[j@jwcxz ~]$ cat /var/log/kdm.log | grep wanted
space: 130984 wanted 131064
space: 130976 wanted 131064
space: 125216 wanted 131064
space: 125208 wanted 131064
space: 130744 wanted 131064
space: 130736 wanted 131064
space: 125248 wanted 131064
space: 125240 wanted 131064
space: 125264 wanted 131064
space: 125256 wanted 131064
space: 125216 wanted 131064
space: 125208 wanted 131064
space: 125264 wanted 131064
space: 125256 wanted 131064

What I find most interesting is the fact that all of these bug reports have been noted in very old versions of Xorg and the Intel driver (all the way back to 2006).  Yet, I'm running the very latest release that's in the testing repository.  I can also say that I have tried the release in the regular repositories and have experienced the same problems on those.

Is this a problem due to the inability to pull more Video RAM?  Can anyone give me any suggestions as to how to make my system stable?

Thanks for your help.

...  This is not the kind of frustration I need two weeks before a paper is due.  I'm thoroughly regretting my being too cheap to buy the nVidia upgrade.


-- jwc
http://jwcxz.com/ | blog
dotman - manage your dotfiles across multiple environments
icsy - an alarm for powernappers

Offline

#2 2008-09-29 07:25:17

Aakko
Member
Registered: 2007-10-23
Posts: 65

Re: xf86-video-intel 2.4.2-1 has "spontaneous" lockups

I also get frequent X lockups with Arch and Intel-driver. I have Dell Latitude C400 with 830M gfx chip. Now I'am running Xubuntu 8.04 as I don't have these lockup issues with it.

Offline

#3 2008-09-29 08:23:51

patroclo7
Member
From: Bassano del Grappa, ITALY
Registered: 2006-01-11
Posts: 915

Re: xf86-video-intel 2.4.2-1 has "spontaneous" lockups

All the problems mentioned are quite well known with the versions of stuff in the testing repo, and everything should be related to DRI (basically TTF has been removed and GEM is not yet there). There is no easy solution (GEM requires hard kernel patches, and anyway the driver for i965 is not ready to use them).

But if you downgrade everything (the server, the drivers mesa, etc) to extra these problems should not happen.


Mortuus in anima, curam gero cutis

Offline

#4 2008-09-29 12:06:35

jwcxz
Member
Registered: 2008-09-23
Posts: 239
Website

Re: xf86-video-intel 2.4.2-1 has "spontaneous" lockups

patroclo7 wrote:

All the problems mentioned are quite well known with the versions of stuff in the testing repo, and everything should be related to DRI (basically TTF has been removed and GEM is not yet there). There is no easy solution (GEM requires hard kernel patches, and anyway the driver for i965 is not ready to use them).

But if you downgrade everything (the server, the drivers mesa, etc) to extra these problems should not happen.

Okay.  Thank you for this information.  Based on (apparently) misinformation that I had read here, I had assumed that the stuff in testing was better.  So, I'll revert back tonight and post back to describe my experiences.


-- jwc
http://jwcxz.com/ | blog
dotman - manage your dotfiles across multiple environments
icsy - an alarm for powernappers

Offline

#5 2008-09-29 14:41:48

dav7
Member
From: Australia
Registered: 2008-02-08
Posts: 674

Re: xf86-video-intel 2.4.2-1 has "spontaneous" lockups

I'm looking forward to this info; I have an Intel 82845G, use the 'testing' repo, and have the same TTM/classic issue, low FPS in glxgears, slow scrolling in Firefox with xcompmgr (forget Compiz, haha), and in general a very similar experience. Gaming? Pah! I forgot about that long ago.

But this may change... big_smile

-dav7


Windows was made for looking at success from a distance through a wall of oversimplicity. Linux removes the wall, so you can just walk up to success and make it your own.
--
Reinventing the wheel is fun. You get to redefine pi.

Offline

#6 2008-09-29 14:44:09

patroclo7
Member
From: Bassano del Grappa, ITALY
Registered: 2006-01-11
Posts: 915

Re: xf86-video-intel 2.4.2-1 has "spontaneous" lockups

Just use the extra repo. The intel driver in testing does not support TTM anymore and is not yet ready for GEM, which moreover is not supported by current kernels. This is not an archlinux problem, but an upstream one. It is discussed also in ubuntu and gentoo forums (by people using the same upstream versions of xorg components).
If on the contrary you want to wait for GEM support, you have to wait for a long period: at least for a 2.6.28 kernel.

Last edited by patroclo7 (2008-09-29 14:47:12)


Mortuus in anima, curam gero cutis

Offline

#7 2008-09-29 19:07:20

freigeist
Member
From: Cologne, Germany
Registered: 2006-07-14
Posts: 191

Re: xf86-video-intel 2.4.2-1 has "spontaneous" lockups

Anyone already dared building the gem enabled kernel and drivers as described at the ubuntu forums?


Elfenbeinturm.cc
a metaphysical space of solitude and sanctity: http://www.elfenbeinturm.cc

Offline

#8 2008-09-29 19:47:33

Dheart
Member
From: Sofia, Bulgaria
Registered: 2006-10-26
Posts: 956

Re: xf86-video-intel 2.4.2-1 has "spontaneous" lockups

Does this mean, that when the gem support is finaly completed we can expect a performance from intel graphics equal to that of a windows machine?

Last edited by Dheart (2008-09-29 19:48:10)


My victim you are meant to be
No, you cannot hide nor flee
You know what I'm looking for
Pleasure your torture, I will endure...

Offline

#9 2008-09-29 21:02:39

patroclo7
Member
From: Bassano del Grappa, ITALY
Registered: 2006-01-11
Posts: 915

Re: xf86-video-intel 2.4.2-1 has "spontaneous" lockups

The promise is of improvements of about +40% FPS in comparison to TTF with INTEL_BATCH (this is what I had read somewhere in an intel website, but I am not able to find the link any more), but I guess that the impact on each supported chipset will be different. I do not know how this compares to windows.

Last edited by patroclo7 (2008-09-29 21:04:01)


Mortuus in anima, curam gero cutis

Offline

#10 2008-09-30 06:19:22

Aakko
Member
Registered: 2007-10-23
Posts: 65

Re: xf86-video-intel 2.4.2-1 has "spontaneous" lockups

Another problem that I have with my Dell and Linux is banding of gradients. My Dell probably only has 6-bit lcd-display and Windows driver enables some dithering and xorg intel driver doesn't. Gradients are much more smoother with windows-driver than they are with xorg driver.

Offline

#11 2008-10-11 01:37:46

anrxc
Member
From: Croatia
Registered: 2008-03-22
Posts: 834
Website

Re: xf86-video-intel 2.4.2-1 has "spontaneous" lockups

jwcxz wrote:

Based on (apparently) misinformation that I had read here, I had assumed that the stuff in testing was better.  So, I'll revert back tonight and post back to describe my experiences.

You can build and install libgl, mesa, intel-dri 7.0.4 and xf86-video-intel 2.4.2 your self (using current/stable xorg-server 1.4.2). Lots of fixes went in there, with some new features.

Changelogs:
http://www.mesa3d.org/
http://gitweb.freedesktop.org/?p=xorg/d … -intel.git


You need to install an RTFM interface.

Offline

Board footer

Powered by FluxBB