When I try to start X with the nvidia driver, the kernel dies. Here's how it looks from /var/log/messages:

Jul 3 11:08:52 3vil Unable to handle kernel NULL pointer dereference at virtual
address 00000000
Jul 3 11:08:52 3vil printing eip:
Jul 3 11:08:52 3vil d107cd26
Jul 3 11:08:52 3vil *pde = 00000000
Jul 3 11:08:52 3vil Oops: 0000 [#1]
Jul 3 11:08:52 3vil PREEMPT
Jul 3 11:08:52 3vil Modules linked in: nvidia w83627hf eeprom i2c_sensor i2c_is
a i2c_viapro i2c_core smbfs ipv6 ntfs
Jul 3 11:08:52 3vil CPU: 0
Jul 3 11:08:52 3vil EIP: 0060:[<d107cd26>] Tainted: P
Jul 3 11:08:52 3vil EFLAGS: 00013296 (2.6.7)
Jul 3 11:08:52 3vil EIP is at os_string_compare+0x13/0x2d [nvidia]
Jul 3 11:08:52 3vil eax: 00000046 ebx: cd190848 ecx: 00000000 edx: c9f428
48
Jul 3 11:08:52 3vil esi: c9f42849 edi: 00000000 ebp: ccd23e30 esp: ccd23e
0c
Jul 3 11:08:52 3vil ds: 007b es: 007b ss: 0068
Jul 3 11:08:52 3vil Process X (pid: 25860, threadinfo=ccd22000 task=cfe60b50)Jul 3 11:08:52 3vil Stack: cd190848 cdb71000 d0ebc25a c9f42848 00000000 c03f958
4 c01340d5 00110000
Jul 3 11:08:52 3vil 00111000 ccd23e60 d0ebe0cb cdb71000 c9f42848 00000000 00001
000 00000000
Jul 3 11:08:52 3vil cdb71000 00000000 c04b4d08 d0911000 ccd23e9c ccd23ea0 d0ec1
9a4 cdb71000
Jul 3 11:08:52 3vil Call Trace:

[Call trace omitted for brevity]
Jul 3 11:08:52 3vil Code: ae 75 08 84 c0 75 f8 31 c0 eb 04 19 c0 0c 01 8b 34 24
8b 7c


This all started when I tried installing kernel 2.6.9-rc1. After installing it, i built the nvidia module, inserted it, started X, and...nothing. Machine totally dead. Thinking that it was just a kernel issue, I restarted into 2.6.7, reinstalled the nvidia module, started X, and...same thing. This happens with versions 6111 and whichever version was before that of the nvidia module.

Here's the really crazy thing: using the nv driver works just fine (though I can't use my dual-head setup). I'm thinking it might be an issue with hardware acceleration, but I have no idea where to start fixing that.

Edit: On a whim, I tried booting into windows, and it died at the boot screen with weird errors. I'm taking that to mean that it's a hardware issue, but that linux and the nv driver are somehow not affected. Anybody have any idea what specifically could be going on, and why nv still works?

It does sound like a problem with the card, as you rightly guessed with the HW Accel. Since the nv driver doesn't use that, where windows and the Nvidia driver does use it.

Time to buy a new motherboard or GFX card. ( Or you might be lucky at it may have worked loose. )