[SOLVED] render error detected, EIR: 0x00000000

cccc · 04-25-2011, 01:17 PM

hi

My gnome desktop from squeeze is completely freezing with these messages:

Code:

Apr 25 18:42:49 sys1 kernel: [ 3030.520012] [drm:i915_hangcheck_elapsed] *ERROR* Hangcheck timer elapsed... GPU hung
Apr 25 18:42:49 sys1 kernel: [ 3030.520025] render error detected, EIR: 0x00000000
Apr 25 18:42:49 sys1 kernel: [ 3030.520042] [drm:i915_do_wait_request] *ERROR* i915_do_wait_request returns -5 (awaiting 28512 at 28511)

BTW where can I find and change Hangcheck timer settings?

mf93 · 04-26-2011, 09:55 PM

It sounds like a video card driver error. Trying reinstalling your graphics drivers.
Also, I've seen this pop up on other Dell(mostly) computers running integrated graphics, especially if they are running there own compiled kernel. The only fix I've found is if you are running a custom kernel and have access to your source. Try this:

Code:

$ locate i915_drv.h

within the kernel source tree(I'm not sure exactly where in the tree off the top of my head) but changing the timeout value there from 75 to 750 solved the problem for other users.

cccc · 04-27-2011, 02:02 PM

Thx a lot, but I'm running defalut not custom kernel.

What I've done and I don't know if it's correct:

First I've download dri-I915-v1.1-20041217.i386.rpm from the Intel website:

http://downloadcenter.intel.com/Deta...are%20Archives

Then I've converted .rpm to a deb file using alien:

Code:

# alien -d dri-I915-v1.1-20041217.i386.rpm
Warning: Skipping conversion of scripts in package dri-I915: postinst prerm
Warning: Use the --scripts parameter to include the scripts.
dri-i915_v1.1-20041218_i386.deb generated

and installed:

Code:

# dpkg -i dri-i915_v1.1-20041218_i386.deb
Selecting previously deselected package dri-i915.
(Reading database ... 89235 files and directories currently installed.)
Unpacking dri-i915 (from dri-i915_v1.1-20041218_i386.deb) ...
Setting up dri-i915 (v1.1-20041218) ...
#

but if I try modprobe then I get this:

Code:

# modprobe  i915
WARNING: All config files need .conf: /etc/modprobe.d/libpisock9, it will be ignored in a future release.

So I've installed libpisock9:

Code:

# apt-get install libpisock9
Reading package lists... Done
Building dependency tree
Reading state information... Done
Suggested packages:
  jpilot pilot-link kpilot gnome-pilot evolution claws-mail sylpheed
The following NEW packages will be installed:
  libpisock9
0 upgraded, 1 newly installed, 0 to remove and 1 not upgraded.
Need to get 282 kB of archives.
After this operation, 512 kB of additional disk space will be used.
Get:1 http://ftp.ch.debian.org/debian/ squeeze/main libpisock9 i386 0.12.5-2 [282 kB]
Fetched 282 kB in 1s (268 kB/s)
Selecting previously deselected package libpisock9.
(Reading database ... 89241 files and directories currently installed.)
Unpacking libpisock9 (from .../libpisock9_0.12.5-2_i386.deb) ...
Setting up libpisock9 (0.12.5-2) ...
# modprobe i915
#

cccc · 04-27-2011, 04:55 PM

Bad news, this problem still exists.

mf93 · 04-27-2011, 07:27 PM

Alright, after doing research, it seems this usually occurs during a heavy multitasking load with integrated graphics. My theory is if our computer is older, it may be running out of ram dedicated to integrated graphics(this wont happen on dedicated cards since they have their own separate memory), and the ram is not freeing up in time to keep the process managing the GUI from timing out, which causes the crash. More memory might fix the issue, or you can obtain the source code of your kernel and manually change the timeout settings. I am personally all for using old computers to their fullest, and kernel compilation is a somewhat painless process(although looking at your credentials it is very likely you already know that)

cccc · 04-29-2011, 06:50 PM

Quote:

Originally Posted by mf93

Alright, after doing research, it seems this usually occurs during a heavy multitasking load with integrated graphics. My theory is if our computer is older, it may be running out of ram dedicated to integrated graphics(this wont happen on dedicated cards since they have their own separate memory), and the ram is not freeing up in time to keep the process managing the GUI from timing out, which causes the crash. More memory might fix the issue, or you can obtain the source code of your kernel and manually change the timeout settings. I am personally all for using old computers to their fullest, and kernel compilation is a somewhat painless process(although looking at your credentials it is very likely you already know that)

Thx, I've upgrade the RAM Memory from 1GB to 2GB, but still doesn't help.
Even I get problems if I try to disable KMS using these entries in /etc/default/grub:

Code:

GRUB_CMDLINE_LINUX_DEFAULT="quiet nomodeset i915.modeset=0"

BTW do you know howto remove or disable i915 and go back to i810 driver?

Code:

# modprobe i810

seems to be OK, no error messages.

cccc · 05-01-2011, 05:37 PM

Quote:

Originally Posted by mf93

within the kernel source tree(I'm not sure exactly where in the tree off the top of my head) but changing the timeout value there from 75 to 750 solved the problem for other users.

I try to build a custom kernel, but howto change the hangcheck timeout value in i915_drv.h:

Code:

/* For hangcheck timer */
#define DRM_I915_HANGCHECK_PERIOD 75 /* in jiffies */
	struct timer_list hangcheck_timer;
	int hangcheck_count;
	uint32_t last_acthd;

mf93 · 05-02-2011, 03:40 PM

change

Code:

#define DRM_I915_HANGCHECK_PERIOD 75 /* in jiffies */

to

Code:

#define DRM_I915_HANGCHECK_PERIOD 750 /* in jiffies */

in other words, change 75 jiffies to 750 jiffies and recompile

cccc · 05-03-2011, 05:36 PM

Thx a lot, I've created a custom kernel and it doesn't help, this error occurs like before.

mf93 · 05-03-2011, 07:05 PM

after some scouring, it seems this is a known unresolved bug in debian as well as almost every other distro. As frustrating as this is for both of us(i hate not being able to resolve issues) there is no solution readily available which I have found. However, keep looking, as you may find something. Also, my help is limited because i do not have a computer running this specific chipset, so i cant get my hands dirty as far as actual tweaking goes. Here is a post which details the attempts of one guy to fix the problem on his computer. http://www.linuxquestions.org/questi...82845g-788563/
I'm really sorry I cannot help you further

edit: i will keep looking though and will let you know if i find anything new

cccc · 06-02-2011, 06:59 PM

I've changed in /etc/apt/sources.list and I've installed these wheezy packages on my squeeze:

Code:

# apt-get install xserver-xorg-video-intel libdrm-intel1 xorg

Now it seems to work well.

mf93 · 06-02-2011, 10:43 PM

Glad to hear it. Just make sure you mark the thread "solved" so other people will know a solution is available. Now on to my own graphics issues...

cccc · 06-09-2011, 06:22 PM

According to:

http://forums.debian.net/viewtopic.p...=63662#p375172

but Option "Shadow" should be "true" (enabled).

another solution for the 82845G video driver problem:

1.) create /etc/X11/xorg.conf

Code:

# Xorg -configure
# cp /root/xorg.conf.new /etc/X11/xorg.conf

2.) put the following in /etc/X11/xorg.conf in Section "Device"

Code:

Section "Device"
Identifier "Card0"
Driver "intel"
Option "Shadow" "true"
Option "DRI" "false"
BoardName "Intel Corporation 82845G/GL[Brookdale-G]/GE Chipset Integrated Graphics Device (rev 01)"
BusID "PCI:0:2:0"
EndSection

Run "lspci | grep VGA" to find out the "BusID" and the "BoardName", for example:

Code:

# lspci | grep VGA
00:02.0 VGA compatible controller: Intel Corporation 82845G/GL[Brookdale-G]/GE Chipset Integrated Graphics Device (rev 01)

cynwulf · 06-10-2011, 02:55 AM

You need to stop copying and pasting other people's solutions from elsewhere: http://forums.debian.net/viewtopic.p...=63662#p375172

This is not the first time I've seen you doing this - it's almost word for word. Give credit to the original poster and provide links.

Also read the comments after that post. According to bug reports, disabling the shadow can cause lock ups on the 8xx chips - which is why a newer kernel is still the better option.

cccc · 06-10-2011, 05:45 AM

Quote:

Originally Posted by Caravel

You need to stop copying and pasting other people's solutions from elsewhere: http://forums.debian.net/viewtopic.p...=63662#p375172

BTW this is a Solution from Chris Wilson (2010-08-26 09:53:16 PDT)
https://bugs.freedesktop.org/show_bug.cgi?id=26345#c113

Quote:

Originally Posted by Caravel

Also read the comments after that post. According to bug reports, disabling the shadow can cause lock ups on the 8xx chips - which is why a newer kernel is still the better option.

Installing a newer kernel from backports or from wheezy repos on my squeeze didn't solve this problem even compile a different custom kernels didn't help.
This is an intel driver issue and not a kernel problem.