When I start X, I get nothing but garbage on the screen. Often but not always this is one and a half rows of light pixels at the top of the screen, with everything else black, which after a few seconds is replaced with purplish-colored columns covering the entire screen. The keyboard stops working, and I can't switch back to a console, use ctrl-alt-backspace, or use ctrl-alt-del. Numlock does work (also maybe caps and scroll lock?), but that's it. I can ssh in, but even when X is killed the keyboard still doesn't work. I'm using xorg-server 1.5.2 and xf86-video-intel-2.5.0. lspci reports the video as 00:02.0 VGA compatible controller: Intel Corporation 82845G/GL[Brookdale-G]/GE Chipset Integrated Graphics Device (rev 03) If I use the NoAccel option, everything works (though obviously, slowly). I will attach the xorg.conf and Xorg.0.log from both working and nonworking tries.
Created attachment 20343 [details] xorg.conf With the exception of NoAccel, this is the output of X -configure. Also, I forgot to mention before, X -configure works and returns control to the console properly, but xorgcfg puts up garbage and there's no way to get control back.
Created attachment 20346 [details] Xorg.0.log from a running server with NoAccel When the server is stopped, only two more lines are added about the keyboard and mouse modules being unloaded.
Created attachment 20347 [details] Xorg.0.log from a running (broken) server without NoAccel This log is from a running server with garbage displayed on the screen. Note that the server tries restarting itself several times (producing garbage each time), and this is the *last* log it produces. I haven't yet managed to get the first log it produces. Also, I forgot to mention, I'm using Gentoo linux, and these are the gentoo packages.
This bug seems a bit similar to bug#17291.
Server works fine with Option "AccelMethod" "XAA" (I guess this is why I never had a problem with 2.1.*) Also, 2.5.1 hasn't fixed it (not that the changelog indicated it would).
What's the kernel version? I've tested recent kernel 2.6.28-rc8 with other git masters of gfx components, it looks interrupt is stuck, so I have to disable DRI, otherwise exa works fine.
2.6.25.19 and 2.6.26.7, I think. I have both of them on my system, and I think I tested with both of them. When I get home tonight (~8 hours from now) I will retest with both and see if I can get a 2.6.27 to test with too. Is there a bug filed on what you said about "it looks interrupt is stuck, so I have to disable DRI" that I can read?
https://bugzilla.novell.com/show_bug.cgi?id=445320 seems to be this
I am now running kernel 2.6.27.10, and it also shows the problem. I also tried using EXA with EXANoComposite, but it was still broken. I will attach some new logs.
Created attachment 21343 [details] Xorg.0.log with XAA and a 2.6.27.10 kernel - everything works
Created attachment 21344 [details] Xorg.0.log with EXA and a 2.6.27.10 kernel - nothing but garbage on monitor
Created attachment 21345 [details] .config for 2.6.27.10 kernel The xorg.conf is unmodified except switching between no acceleration, XAA, and EXA.
Is there any more information I can give to help this get resolved?
I didn't see this mentioned, but adding: Option "DRI" "false" to the Device section fixes the issue for me. I have AccelMethod as EXA and NoAccel left as default. Moving windows is not choppy like when NoAccel is "true". Just noticed Comment #6 mentions disabling DRI. I'll leave this in simply because it shows the xorg.conf option.
EXA is going away, according to http://intellinuxgraphics.org/2009Q1.html, so this doesn't matter. UXA doesn't have this problem, anyway. Making a guess that this resolves as wontfix?
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.