From Bugzilla Helper: User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.1) Gecko/20020827 Description of problem: When X is started by Anaconda the machine hangs, requiring a hard reset to continue. Version-Release number of selected component (if applicable): How reproducible: Always Steps to Reproduce: 1. Boot from Psyche CD 1 2. Press return to perform a graphical install Actual Results: Video card is detected as GeForce 2 MX. A couple of seconds later the screen goes black, and the machine hangs completely. Expected Results: Working X Additional info: Hardware: Asus A7N266VM; the closed-source X driver from NVidia works perfectly on Redhat 7.3 (as it's being used while writing this bug report :-). This is the exact same problem as I have previously reported in bug 70022, and it looks the same as bug 72721 as well.
I have tried swapping nv_drv.o for the one available in the official binaries at xfree86.org, but the behavior is the exact same - a hard lockup. It seems as if the log files contain no useful information; I guess they get truncated whenever the machine locks up.
Created attachment 79146 [details] X log file with working driver from current CVS
I have tried with a build of nv_drv.o made by Mark Vojkovich from the current XFree86 CVS and it appears to work perfectly. Of course this doesn't help with the installer, but hopefully I'm able to get things going in text mode and change the driver from there. I have attached a log file made with the working driver.
I've got XFree86 CVS RPM packages on ftp://people.redhat.com/mharris which you can test if you like. Since you've indicated that this is fixed in CVS, I'm closing the bug report as RAWHIDE, since these CVS RPM's are going into RAWHIDE this week.
I can confirm that the CVS build solves the problem - unfortunately introducing a new bug along the way, but I guess you can't win 'em all :-) The new bug is that the Mozilla throbber and scrollbars become completely garbled. Should I report this as a bug on Rawhide, or just wait for a new build?
Thanks for testing and confirming the problem is fixed in the CVS builds. I've noticed your new bug report also, and will work with Mark V to try and find a solution.