Re: [SLE] Re: NVIDIA once again

On Sun, 2004-03-07 at 06:08, John Pettigrew wrote:

In a previous message, Terry Eck wrote:

<snipped>

Now I've found the problem - I was looking at the wrong set of instructions for SuSE 8.2. Strangely, there are 2 sets, only one of which mentions the need for the "make cloneconfig && make dep" stage. Having done this, the driver reported a successful build and install.

Thats because you're still following the wrong instructions from the sounds of it :).

However, trying to use this driver in any way (e.g. running sax2 or starting an X session) caused the screen to corrupt (a few blue lines in the bottom inch of the screen) and the computer to crash terminally - unlike any other crash I've seen, it actually required me to reset the PC and fsck to catch corruptions.

WTF is going on here?

Do the proper install for 8.2 may help. Did you perchance do "sh nvidia-installer###.sh --extract-only"? This is fine for 8.1 where the newer installer does not work. Doing sh nvidia-installer###.sh (### conforms to the other bits that go with the drier) will installer the installer, which can then be launched via nvidia-installer. I would try nvidia-installer -f which is for it to force an update. If you still have problems with corruption (the software sort, not the SCO sort!) then "lspci" will help. Dmesg will be hard if you're getting hard locks. Was having issues setting up my brothers computer, from 5000 miles away...3D just kept saying disabled under /proc/driver/nvidia/agp/status (can "cat" this, cat /proc/driver/nvidia/agp/status) and was frustrating as hell, then under "dmesg" agpgart was stating that the SiS host 0746 was unsupported and to try a switch (cannot remember off hand), upshot of this...It works sometimes, but NvAGP, the nvidia driver does not at all. So, try again with the install, the way SuSE 8.2 is meant to be done, then if you have issues we'll take it from there...Right now, I am writing to SiS asking for better Linux support, they can surely spare an engineer to contribute patches and information on their chipsets... Matt