Red Hat Bugzilla – Bug 26453
Testing X configuration fails erroneously
Last modified: 2007-04-18 12:31:16 EDT
Hardware: Dual P-II 400, on QDI BrilliantIV (440BX, BIOS version 3 beta) Diamond V770 Ultra (TNT2 ultra chipset) Samsung SL900+ (using probed DDC syncs) 256megs RAM 6 Gig Maxtor over UDMA/33 on /dev/hdc USB mouse (Razer Boomslang 2K) Install parameters: Custom install, all packages. Automatic partitioning X selected at 1280x1024, 24 bit. Dual-booting with Win2K (on /dev/hda1, whole disk) Testing the X config fails with "Explicit kill or server shutdown". However, the resulting XF86Config-4 file is perfect, and startx (or booting to init 5) works flawlessly post-install.
Assigning to a developer.
I should also mention, in case it is relevant: The V770 has 32 megs video RAM, but anaconda detected it as 16. I changed it to 32 from the dropdown. I will hammer on this with some other settings when I have free time, and see what the reproducable envelope is.
Can you go to VC2, run ddcprobe, and post the output? That way we can tell how much memory the card thinks it has.
We (Red Hat) should really try to resolve this before next release.
You might also want to look at /tmp/XF86Config-4 (or whatever it's called after clicking on the test button, to see what the differences are between that and the installed Xconfigurator config file.
We think this has been fixed in the latest internal builds. There was a problem in the way we were handling memory. Basically, testing X caused a spike in memory usage, and the kernel was killing the X process because it thought it had run out of memory. Please reopen if the problem persists.