Linux Debian support - agpgart drm hangs and then so does make-kpkg

This is Interesting: Free IT Magazines  
Home > Archive > Linux Debian support > July 2007 > agpgart drm hangs and then so does make-kpkg





You are viewing an archived Text-only version of the thread. To view this thread in it's original format and/or if you want to reply to this thread please [click here]

Author agpgart drm hangs and then so does make-kpkg
Geico Caveman

2007-06-29, 1:14 am

Hello

I finally had my laptop's system board replaced today and found that I
cannot seem to get module i810 (the new graphics chip is Intel Mobile
945GM/GMS/940 GML) to start kdm because /dev/agpgart and /dev/dri are not
created. I have a Fill_in_dev error in dmesg.

So, I am forced to work with the vesa driver. That is so slow (slower than
even my old laptop used to be), that I think that there is a serious kernel
problem. But this is the debian stock kernel 2.6.18-686.

Anyway, I booted with a kubuntu live cd (which had no problems) and copied
the kernel over and tried to compile it. Now make-kpkg hangs at :

debian/linux-headers-2.6.20.5/DEBIAN/postinst
cp -pf debian/control debian/control.dist
k=`find /usr/src/linux-2.6.20.5/debian/linux-headers-2.6.20.5 -type f | (
while read i; do \
if file -b $i | egrep -q "^ELF.*executable.*dynamically linked" ;
then \
j="$j $i"; \
fi; \
done; echo $j; )`; test -z "$k" || dpkg-shlibdeps $k; \
test -n "$k" || PERL -pli~ -e 's/\$\{shlibs:Depends\}\,?//g'
debian/control

It does not matter how much time I give it - 2 hours and waiting.

What is now so messed up with Intel graphics drivers ? I had been using
nvidia for about 2 months, and gave up on that graphics chip because of
persistent black window bug and a hibernate problem that the nvidia team
seemed incapable / unwilling to fix (1 year and counting - these bugs).

Any suggestions would be welcome. I really (and I repeat really) do not
want to overwrite the debian installation with the kubuntu one. And yes, I
AM using the SAME xorg.conf in both kubuntu and debian (copied it over).

Thanks.

--
Posted via a free Usenet account from http://www.teranews.com

Geico Caveman

2007-07-02, 7:14 pm

AJackson wrote:

> What exactly is the error lines?
> You might want to look if there is a bug reported on this. And then
> file a bug report to Debian about it.


I have found similar bug reports that seemed to relate to the separate
loading of agpgart driver.

Anyways, I figured it out - I had blacklisted intel-agp driver during my use
of the nvidia hardware. Once I fixed that, the problem went away. I am
using the i915 driver :

i915 17600 2
drm 61332 3 i915


However, I have now uncovered other (more serious) problems.

The system now seems to run at higher temperatures at times (as high as
101C or so - it used to be up to 59C max with nvidia, which drives the fan
a lot harder. This is not a fake temperature as I can feel the base of the
laptop becoming fairly warm. The log files indicate sustained periods when
the temperatures are close to 100C or so.

When the fan runs a lot harder, everything seems to slow down (as
the CPU slows down as well), sometimes so slow that it seems to be even
slower than my old single CPU Celeron laptop. I specifically checked for
memory usage during such periods of slowdown - there is no noticeable
change over the more normal periods - so it is not as if the system is
running out of RAM space and swapping (swap space usage is consistently
zero or close to it).

Another thing is that the slow down persists even when I am in the console
(even after stopping kdm) - which is another indication that the problem is
not related to the nature of the graphics chip.

The craziest thing is that when I run top, I find that some processes by
root, specifically, hald and modprobe (I kid you not) are taking up as much
as 30% of the CPU cycles each. And they seem to run throughout the slowdown
period during which the system slows to a crawl, and the fan is going
gangbusters.

Now, I thought that modprobe was run like once when the system booted /
came out of sleep, etc. The fact that it runs for this long and this
correspondence to a slow down is very very puzzling and troublesome.


>
> It also could be that another module is using the device, so the agp-
> module can't take contoll over the i945. Are intelfb loaded?
> Try command "modinfo i810"
> Should you not use module i915 ?


Here are my loaded *intel* modules :

$ lsmod | grep "intel"
snd_hda_intel 17332 2
snd_hda_codec 137856 1 snd_hda_intel
snd_pcm 68676 3 snd_hda_intel,snd_hda_codec,snd_pcm_oss
snd 47012 12
snd_hda_intel,snd_hda_codec,snd_pcm_oss,
snd_pcm,snd_mixer_oss,snd_seq_oss,snd_ra
wmidi,snd_seq,snd_timer,snd_seq_device
intel_agp 21148 1
agpgart 29896 3 drm,intel_agp
snd_page_alloc 9640 2 snd_hda_intel,snd_pcm


No intelfb.

>
>
> You shouldn't copy the kernel over between Ubuntu and Debian. What
> did you mean by copy?


I copied over the xorg.conf file and was about to start compiling a kernel
using the copied over .config file from the Ubuntu live CD. In any case,
the reason for doing that is moot since I have solved the original problem
and uncovered a few more that appear to be worse

Thanks.
Sponsored Links






Free braindumps | Software forum | Database administration forum

Copyright 2003 - 2008 webservertalk.com