intel graphics - X freezing frequently

[asturm]

I think that's a different problem altogether, I don't have native Firefox installed though, wine-firefox renders the site as well as any other browser on my system (opera, konq, rekonq, arora).

Concerning the Intel driver, I now switched back and forth 2.11 and 9999 a few times as latest git revision brakes X again (I guess it needs several additional live X deps), thankfully I had quickpkgd the last working state before. But 2.11 also ran flawlessly that time, so maybe your best bet would be to move on to latest git-sources-2.6.34 and see if it fixes the freezes for you.

[Yamakuzure]

Well, not everything is fine here. (But this cricket website renders fine with FF 3.6.3)

using xrandr crashes xorg-server, which then restarts.
full screen opengl games crash the whole system after some minutes. At least pressing the power button shuts the system cleanly down.

I neither use xrandr nor full screen opengl on this machine, that's why I've never witnessed any crashes, sorry.

[rahulthewall]

I am now annoyed enough to add this for the time being:

Code: Select all

googly rahul # cat /etc/portage/package.mask 
## Xorg Mask
>=x11-base/xorg-server-1.8.0
>=x11-video/xf86-video-intel-2.11.0

EDIT: Downgrading is "one hell" of a process. Someone shoot me.

[Yamakuzure]

Someone shoot me.

ok...

Code: Select all

   o                              \O_ Arrgh!!
  <\==-   -   - -   -  -  - --- __/
  / \                             \

Now, I do not think that this is a kernel problem. For me it looks like the new intel drivers are not fully upgraded to work with xorg-server 1.8*. Unfortunately this means we have to wait for the next driver version.
Or at least I am hoping that it is the intel drivers only to be blamed, because if it is the combination of mesa + libdrm + intel driver, this is bound to take while.

[rahulthewall]

Or at least I am hoping that it is the intel drivers only to be blamed, because if it is the combination of mesa + libdrm + intel driver, this is bound to take while.

Cool ASCII art.

And I sincerely hope this is the case. Because right now my system is anything but usable.

[Bakterie]

Now, I do not think that this is a kernel problem. For me it looks like the new intel drivers are not fully upgraded to work with xorg-server 1.8*.

I'm having this problem with xorg-server 1.7.6 and intel drivers 2.9.1. Any kernel > 2.6.31 causes the problem. Which are the "new" intel drivers?

[jsevilleja]

For what I've read on this thread, it seems like you're having this bug: https://bugzilla.kernel.org/show_bug.cgi?id=15417
There's a temporal solution, but is not valid for 2.6.33.4 and up. I think that the bug depends on the vendor of the graphics card. My laptop, with an GMA 4500MHD, is affected (manufactured by Packard Bell), my girlfriend, with the same card, but manufactured by Dell, isn't.

[asturm]

Yes, I stumbled over that one already. The description comes closest to my experience with xrandr.

[Sh4doW]

From my experience there's something not working right with xorg-server-1.8.x and xf86-video-intel-2.11.0 with GM965, no matter the kernel version (tried gentoo-sources 2.6.33 and 2.6.34 on x86).

Xrandr failing to change resolution is a show-stopper for me, but besides that, haven't really noticed any other mentioned anomalies.

I'm staying on 1.7.x series for now. Will report back if there are any changes on my end (doing ~x86 updates weekly).

[rahulthewall]

From my experience there's something not working right with xorg-server-1.8.x and xf86-video-intel-2.11.0 with GM965, no matter the kernel version (tried gentoo-sources 2.6.33 and 2.6.34 on x86).

Xrandr failing to change resolution is a show-stopper for me, but besides that, haven't really noticed any other mentioned anomalies.

I'm staying on 1.7.x series for now. Will report back if there are any changes on my end (doing ~x86 updates weekly).

I am back on 1.7.x for now as well - I think I will now update when 1.9.1 is out.

[jbouzan]

Did anyone find a fix besides downgrading? I've had a lot of trouble getting any graphical display with this laptop, and I was never able to get a working config except with xorg-server 1.8. But it's started having frequent freezes on waking from sleep or when opening software (practically anything, I've seen it with konqueror and firefox and kate and konsole). Not sure what changed last week to cause this.

[Gh0str1d3r]

Did anyone find a fix besides downgrading? I've had a lot of trouble getting any graphical display with this laptop, and I was never able to get a working config except with xorg-server 1.8. But it's started having frequent freezes on waking from sleep or when opening software (practically anything, I've seen it with konqueror and firefox and kate and konsole). Not sure what changed last week to cause this.

I am still experiencing freezes, even though they are very rare (at last it ran for 1.5 weeks 8 hours a day without problems, then it froze again). I am now using kernel-2.6.33-gentoo-r2, xorg-server-1.8.1 and xf86-video-intel-2.11.0, KDE-4.4.3 with compositing. I don't have any graphic bugs, never had them. Only connecting a second monitor requires some repeating of choosing the correct resolution in the KDE display menu, and the second monitor must not be connected during boot.

I am wondering if the graphic bugs are really the same problem as the freezes. But anyway, both should be solved soon!

[asturm]

I recently had one freeze again where everything but the cursor stalled. That after weeks of unproblematic work. Please look into your kdm.log for an intel_bufmgr_gem.c error after such freezes, there was one report in the according bugzilla entry: https://bugs.gentoo.org/show_bug.cgi?id=318743

If this error is reproduced among other users we should report this to the kernel bugzilla.

EDIT: I'm now 'back' on live ebuilds for server, intel driver and some more deps. Let's see where that gets me. xrandr now completely throws up, but it couldn't get much worse.

[asturm]

xorg-server-9999 freezing too.

It happened now for the second time when trying to access the (composited) hidden kde4 taskbar.

EDIT: Since X live didn't bring any success, I'm now going to try a 2.6.34 kernel with 2.6.35 drm subsystem (made it build, testing now).

[asturm]

Haven't seen a freeze in a while, but still that recent update in libdrm master makes me hopeful it might has been fixed now:

Code: Select all

 include/drm/i915_drm.h    |    5 ++++-
 intel/intel_bufmgr.c      |   13 +++++++++++++
 intel/intel_bufmgr.h      |    3 +++
 intel/intel_bufmgr_gem.c  |   34 +++++++++++++++++++++++++++-------
 intel/intel_bufmgr_priv.h |    7 +++++++
 5 files changed, 54 insertions(+), 8 deletions(-)

Since the freeze has been linked with intel_bufmgr in the according bug report...

[Gh0str1d3r]

I still have freezes about every 2 weeks. Even though this is not too problematic for normal work, it makes the computer unusable for presentation. So lets hope for a fast release of the new libdrm!

[jsevilleja]

Haven't seen a freeze in a while, but still that recent update in libdrm master makes me hopeful it might has been fixed now:

Code: Select all

 include/drm/i915_drm.h    |    5 ++++-
 intel/intel_bufmgr.c      |   13 +++++++++++++
 intel/intel_bufmgr.h      |    3 +++
 intel/intel_bufmgr_gem.c  |   34 +++++++++++++++++++++++++++-------
 intel/intel_bufmgr_priv.h |    7 +++++++
 5 files changed, 54 insertions(+), 8 deletions(-)

Since the freeze has been linked with intel_bufmgr in the according bug report...

I'm testing 2.6.35-rc2 right now. Everything seems normal to me, but time will say

[asturm]

Froze again (2.6.35 drm system, latest libdrm). Oh well... anyway, big changes to libdrm's intel_bufmgr again, so maybe it was fixed this time...

[Tolstoi]

I ran into this problem as well. At the moment everything looks fine, but you can never tell......

Funny message I've never seen when compiling xf86-video-intel at the end of an emerge:

Code: Select all

* Messages for package x11-drivers/xf86-video-intel-2.11.0:

 * getfilevar_noexec requires 2 variables, with the second a valid file.
 *    getfilevar_noexec <VARIABLE> <CONFIGFILE>
 * This driver requires KMS support in your kernel
 *   Device Drivers --->
 *     Graphics support --->
 *       Direct Rendering Manager (XFree86 4.1.0 and higher DRI support)  --->
 *       <*>   Intel 830M, 845G, 852GM, 855GM, 865G (i915 driver)  --->
 *               i915 driver
 *       [*]       Enable modesetting on intel by default
>>> Auto-cleaning packages...

Is that just some kind of reminder or a hint that KMS was not found. The asterisks before the getfilevar lines are red.

[Gh0str1d3r]

today came an update of libdrm, lets see if they fixed the bug.

@Tolstoi: That's a normal reminder that you need to enable KMS. I guess it is red because there are so many emerge remarks that most people only read the red ones carefully.

[asturm]

Don't know how far it is up to the latest commits, at least since my last update of libdrm-9999 no freeze has happened.

[brynstntn]

The developers at Red Hat seemed to have found a workaround. They disabled page flipping. I've tested tuxracer on a fedora live cd, and had no crashes.

http://lists.fedoraproject.org/pipermai ... 35429.html

[Gh0str1d3r]

since our freezes occur even with xf86-video-intel < 2.11 page flipping is most likely not the (only) reason for the problem we are talking about.

I will try the libdrm-9999 package and report if it is stable or not.

[Tolstoi]

The freezes I had suddenly disappeared. The .9999 driver doesn't work at all for me.

[Gh0str1d3r]

libdrm-9999 (version 18.06.2010) still freezes