so here's what i see in syslog when i try to end session:

Aug 19 06:00:51 localhost kdm[3252]: X server for display :0 terminated unexpectedly
Aug 19 06:00:51 localhost kdm[3252]: Unable to fire up local display :0; disabling.

blinking text cursor in upper left corner of the screen.....

ctrl + alt + F5 or F6

as root:

killall -HUP kdm

Find somewhere to put that after session logout.

I haven't had time too look into it but I get it on intel with Kde 4's kdm randomly.

It's constant on Kde 3.5.10's kdm.

The culprit is probably xorg-video-intel / Mesa

Ubuntu 9.10 rc3 has it.


I'd recomend
a. kernel 2.6.31 rc5
b. https://sourceforge.net/project/down...-1.5.1-src.tgz may need git/cvs
c. http://www.mesa3d.org/beta/MesaLib-7...0090811.tar.gz
or if you know git pull it from there.
d. libdrm 2.4.12 or later


Instead of us all building these on our own we should set up a sourceforge site or something.

I think I'll give Slack13 a run on my laptop which has an old ATI chip and see how it goes - 12.2 is a bit glitchy on that - before I dive back into trying to get the Intel driver to work. It would be nice to see packages that can make it work in /extra, maybe under an Intel vid section. Not exactly ideal either way but, then again, Slackware is reputedly for those who like a bit of extra work.

Anyone had any better results with the 2.8.1 Intel vid driver build that's in /extra of Slack 13?

explain "find somewhere to put that"..... you mean as a logout script or something?

That will fix the hanging at the console described earlier...

2.6.31 rc9 is out so you probably want to go with that version of the kernel. But as always check the kernel ChangeLogs.

Mesa CVS/GIT also.

http://lwn.net/Articles/348883/ Intel release make news here.

With all the patches in place Unreal04 runs about 80% of it's Windows version.
But games like Warsow, Nexiuz run smooth as silk.

hmm, Sasha pointed me in this direction in response to some problems I was having ... I've gone through the entire thread, as well as the older one about the same topic, and Im pretty sure Im no closer to the solution than I was before reading this ... Only because Im not on the same level of expertise as the people discussing stuff in this thread ...

anyways, long story short, could some one please point out if the following symptoms are related to the topic being discussed in this thread, and if yes, how bout some pointers on where I begin, to get started sorting this mess out -

System is a pretty old one, Intel 845G chipset motherboard with whatever on-board graphics (dont remember right now), Pentium D 2.4GHz, 768MB DDR RAM with 8MB shared for the on board graphics.

Slack 12.2 - fresh install, no changes to anything, COMPLETE installation - startx leads to the nice flashy GUI, but ctrl+alt+fx will NOT bring me back to any of the consoles, nor will ctrl+alt+bkspc terminate X!

I was pretty sure that it was a graphics driver related issue, tried to google to figure out a solution, noticed 13 had just been released, so got that and reinstalled over 12.2. NOT upgraded, full format and reinstall!

Slack 13 - fresh install, no changes to anything, COMPLETE installation - startx leads to the even nicer and flashier GUI, and ctrl+alt+fx/bkspc all work exactly as expected. Problem is that the system 'hangs' after about 30 minutes or so! Mouse still works, num-lock still works, so its not completely frozen, but absolutely nothing else works, only way to move on is to hit reset and reboot!

I never used 12.2 long enough to know if the same hanging problems existed there too!

Im not scared of the CLI, quite like using it, BUT my level of experience with Linux is not too great, so if this IS a related problem, and if I can help you guys out, while getting my own system to work, you will need to do a little bit of hand-holding when it comes to information that I need to get out of my system and provide to you.

Oh also, I tried putting in an ancient nVidia Riva TNT2 AGP, just to see if the problem would go away, meaning it was Intel causing it ... The system went NUTS during bootup with pages and pages of errors before it 'froze'! Just remember that a lot of it had to do with udev! The system booted up again once I pulled the nVidia card out, I am pretty sure that inserting a new AGP is NOT an 'illegal' thing to do, meaning I dont think a simple swap ought have caused THAT many errors, but I could be absolutely wrong, of course!

My thread is located at http://www.linuxquestions.org/questi...13-gui-752119/ ... but be warned, there are a few posts that are totally off-topic in that thread (courtesy, me!) So, serious folks who cannot waste time, probably a good idea to ignore that thread

Tried all the alternate drivers, anything below 2.8.1 hangs with just the mouse shown, doesnt go any further!

2.8.1 hung once after about 18 minutes or so, I guess ... This is round 2, trying to get this post out before it crashes!

EDIT: It crashed seconds after I got that post out!

Which kernel? It looks like you're using the stock one - how about trying the later model from /extra?

I was using the stock kernel. Though, i did finally download the 2.6.30.5 kernel yesterday! I had time only to do one kernel compile, and I tried to get the TNT2 working with it, no luck! When I enable rivafb support in the kernel, booting up with the new one manages to give me a kernel panic - vfs ... unable to mount root on any fs ... unknown partition type ... and so on! Booting the same new kernel with the onboard graphics leads to a usable terminal, but startx wont work, probably because I did not include the appropriate drivers! Did not have time to try the alternate intel drivers with this new kernel, will report back once I try that. What Im VERY confused about is WHY inserting a TNT2 card should lead to a kernel panic which lists the Filesystem as a problem!!!

As for kernel compiling itself, truth betold, the last time I used Linux and did any form of 'successful' kernel compiles was with Slack 10.1 ... and if Im not mistaken, 2.6.10 something kernel!!! The config options have changed quite a bit since then, so I am definitely not too very sure that I configured the new kernel correctly! Is there someway to use the 2.6.29.6-hugesmp (stock)'s .config file, add rivafb support to it, and then compile the 2.6.30.5 with it? Oh and also would need to know WHERE to look for the original .config file, coz I did mrproper a few times with the stock kernel, so the original is probably long gone from the src folder!

The config files used to compile the stock kernels are all in your /boot directory.
You can copy one of them as a model to /usr/src/linux/.config and do a "make menuconfig" to just set the options you need to change.

will take a look-see at the config files ... but heres something interesting ... running Seamonkey now, on the stock kernel with the 2.8.1 driver ... disabled as much eye candy stuff as I could from KDE ... so far, the system seems to be fine ...

Wondering if there is some kind of 'reaction' that begins when the mouse pointer is moved over a button? Every single crash that happened when I was using firefox happened at the exact moment that I moved the mouse pointer over the post reply button! So far so good ...

moved the mouse between the terminal emulator and the seamonkey windows, opened side by side, hung again!

In 13 /testing/packages there's a 2.6.30.5 pre-compiled for you:

http://slackware.osuosl.org/slackwar...inux-2.6.30.5/

But yes, you can snag a config file from /boot and use the makefile option that only prompts for novel config options that are in the new kernel source and accepts the old ones as default. This way it's easy to build a copy of the stock Slack kernels from a new kernel source.

But I'm most interested that people give the /testing a kernel a go.

If I can find another HD that will fit in my machine I may give it a go over the w/end but I'm loath to mess around with the working 12.2 I've got now.

Edit: in /usr/src/linux do a `make help' and you'll see the options:
`make oldconfig' will do the job if you copy a copy a config file from /boot to /usr/src/[new_kernel_src]/.config

But, as I said, give the /testing kernel a go - that's going to be a quick fix.