Linux 9626 beta driver dies on X startup if DFP-0 is connected to 6800U when booting

[LordHavoc]

Summary of the problem:
X locks up at startup when using 9626 beta or 9640 NVPerf beta drivers, unless I physically unplug DFP-0 during the BIOS stage of bootup, then I can happily reconnect it after the BIOS stage and everything works. This problem does not occur when using 87xx drivers, and I have switched back and forth several times and experimented extensively with xorg.conf options (including UseEDID FALSE, CustomEDID with edid files dumped from my monitors, and also the ConnectedMonitor option).

Relevant Hardware:
2x Viewsonic VP2030b 20.1" LCDs (1600x1200x60hz), BFG Geforce 6800 ULTRA OC 256MB AGP, Asus K8V (VIA K8T800 chipset) motherboard.

Explanation of my monitor setup:
I have one monitor hooked up as DFP-0 and the other as CRT-1, one is on DVI-D, the other on DVI-I which gets misdetected as DVI-A due to some (latency?) issue with this card and these monitors, but I prefer that behavior because the right monitor can't handle 1600x1200x60hz 161mhz non-reduced-blanking signals from the external TMDS encoder (it gets flickering ghosting artifacts to the right of any horizontal dark lines where the pixels are offset by a subpixel offset, it's weird), the internal TMDS encoder driving DFP-0 delivers a 1600x1200x60hz 140mhz reduced blanking signal which works perfectly.

Notes:
The attached bug report log is when I have it working, since I can't get any log out of it when it's not working - the X server leaves my displays blank (except for sometimes having a cursor in the top left), and there's nothing interesting in the Xorg.0.log, even if I use startx -- -verbose 5 -logverbose 5.

Also when it locks up like that I have to physically power off the machine because although ctrl-alt-del causes a reboot the reboot process never finishes (nvidia driver fails to unload).

[LordHavoc]

Just wanted to say, the 9746 driver fixed this problem.