Intel Pushes Out Its 2.12 X.Org Driver To RC State

[DebianAroundParis]

Please Intel get rid of this:
[drm:i915_wait_request] *ERROR* i915_wait_request returns...
I am tired of all those reboots!

[Linuxhippy]

If it suddenly started and downgrading doesn't help, have you ever considered your hardware to be broken?

Please Intel get rid of this:
[drm:i915_wait_request] *ERROR* i915_wait_request returns...
I am tired of all those reboots!

That means your GPU is hung. So instead of spreading bad publicity in a forum not read by intel employees, you could do:
1. Install kernel 2.6.34 (needed for debugging)
2. install intel 2.11.901 + libdrm-??.21
3. When a hang occurs, log in via ssh save the GPU-state as well as the last batchbuffer. File a bug at freedesktop's bugzilla, and append those files:

After a freeze, can you grab a copy of dmesg, Xorg.log, intel_reg_dump,
and /sys/kernel/debug/dri/0/i915_error_state. From my perspective, the
last contains the most interesting information (the batchbuffer executing
at the time of a gpu hang). If there is nothing in any of those, echo t >
/proc/sysrq-trigger and look for the stacktraces in dmesg.

I guess they would be more than happy about feedback like this.

- Clemens

[Xake]

If it suddenly started and downgrading doesn't help, have you ever considered your hardware to be broken?

That means your GPU is hung. So instead of spreading bad publicity in a forum not read by intel employees, you could do:
1. Install kernel 2.6.34 (needed for debugging)
2. install intel 2.11.901 + libdrm-??.21
3. When a hang occurs, log in via ssh save the GPU-state as well as the last batchbuffer. File a bug at freedesktop's bugzilla, and append those files:

Nice, now could you get someone to actually look at Intels bugreports?
The ones I find when searching for i915_wait_request in bugs.freedesktop.org tells the same story as most of the bugreports I have done about Intel and their graphic stack: If you are lucky you get a reply, if you are lucky you get a reply asking for more information. If you are really lucky you will get more of the same question about three months apart.
I have at least two open Intel bugreports currenntly, I am following two others and would have had more if I actually had time to track.
The first is about the wrong resolution being set for S-Video out upon boot (should be a with of 640, got 848), filed nine months ago. They provided a patch in the bugreport that set the right resolution at least 2 kernel releases ago, but still 2.6.35-rc4 does not set the correct resolution. Poingting it out in the bug about two months ago, no answear in over four months.
The second is about an issue with the resolution being set having the wrong clock for some resolution. The first time I mentioned the problem (but still did not new what coused it nor did I get any help try to figure it out) was in the former bugreport. Since then they manage to break all modes for NTSC-M and PAL here on my i945 S-Video out. Filed a new bug for this over one month ago with all information asked for in their wiki, still no answear. And this is broken out of the box on Fedora13, I only need a LiveUSB of F13 to reproduce, no logging in or fiddeling needed.

Beside those I have seen Xv hanging the whole X-server (being tracked on fedora/redhat, there at least some maintainers who help try to figure out stuff). I have also some other issues that I have filed on redhat and waiting for input.

I guess they would be more than happy about feedback like this.

- Clemens

Yeah, I guess they are. As long as they can ignore it and instead try to optimize the codebase instead of first getting it to work for all their costumers (yes, they are employed by intel, and I have bougth intel hardware, I am a costumer who wish to be able to use Linux on my boxes, not just windows).

And yeah, I am a little bit irritated since Fedora13 is broken out of the box on one of my regular laptops (a i855 machine which hangs and have all "shadowed" buttons white and some other graphical artifacts) and on a little computer wich should only be connected to a TV for various things (a box using a i945GM chipset and S-Video, works with UMS, totally broken with KMS).

[Xake]

Oh, and yeah. I really hate those 1 minute lock on posts here. Could you at least make it 5 or maybe 10 minutes? Or remove edit completely?

[DebianAroundParis]

If it suddenly started and downgrading doesn't help, have you ever considered your hardware to be broken?




That means your GPU is hung. So instead of spreading bad publicity in a forum not read by intel employees, you could do:
1. Install kernel 2.6.34 (needed for debugging)
2. install intel 2.11.901 + libdrm-??.21
3. When a hang occurs, log in via ssh save the GPU-state as well as the last batchbuffer. File a bug at freedesktop's bugzilla, and append those files:



I guess they would be more than happy about feedback like this.

- Clemens

Weeks ago I did a search and I saw my problem already reported all over the web for months.
Lately I noticed that this bug seems to be triggered by a combination of VLC and KDE.

[darkbasic]

Intel driver is a bad joke, what performance improvements they are talking about when some basic functionality causes X server hangs, when OpenGL performance is ... the word "performance" doesn't apply well to Linux Intel GPPU drivers.

Quote

Nice, now could you get someone to actually look at Intels bugreports?

Quote++

I hate the intel driver, next laptop will probably have a radeon.

[colo]

I threw out my G45 yesterday (well, I actually didn't, as I failed desoldering the actual silicon ), and got a cheap, used Radeon 4350 instead. Could not stand the slow rendering with intel (2.11 still, though) any more, so it just had to go. The difference is _huge_, and I'm not looking back at the moment.

[marco]

i915 hang CPU

I'm un unhappy owner of a 945GM intel video card.

Unhappy 'cause all I tried in compiling git master branch for mesa, xf86-vide-intel led to hang my GPU since 2.6.33 kernel.

I'm using gnome-shell from git and every ten minutes my Xorg crash saying this:
[drm:i915_hangcheck_elapsed] *ERROR* Hangcheck timer elapsed... GPU hung
I think it's because clutter uses large texture and their are GPU intensive

I googled everything until yesterday, then I found this patch:

https://patchwork.kernel.org/patch/104542/

I rebuild my 2.6.35-rc3 kernel from git appending the patch and... VOILA... all the hangs have gone.

Could you guys (owners of intel hardware) confirm the patch resolve GPU hang?

Now i'm an happy owner of Intel card.

Bye
Marco