Problem with DSS DVI output on DM3730

OMAP forum:

We are proving out the DVI output on a DM3730 board that we've designed, and we're seeing something strange: Our output to the video window is always red.  We see the penguin show up in the corner, and text/graphics show up, but the red channel is always on for all pixels on the screen.

Studying our schematic and comparing with the Beagleboard-xm schematic, it appears that we're hooking everything up the same way, with just some subtle differences:

We desire to output 1280x720 resolution (720p), and that means we need to use the alternate configuration for the DSS output pins, as mentioned in the DM3730 TRM:

On Page 1564, it states:

The DISPC_DATA_LCD[5:0] data multiplexed with the DSI signals on dss_data[5:0] pads is limited to up to 60 MHz pixel clock frequency.
If the parallel 18/24-bit interface in bypass mode with a pixel clock above 60 MHz is required, the DISPC_DATA_LCD[5:0] multiplexed on
dss_data[23:18] pads, and DISPC_DATA_LCD[23:18] multiplexed on sys_boot pads must be used.

So, we've mapped DISPC_DATA_LCD[5:0] to dss_data[23:18] pads and routed DISPC_DATA_LCD[23:18] to be multiplexed with the sys_boot pads.  

Our sys_boot pads are all pulled up to 1.8 V through 4.7 K-ohm resistors.  I'm suspicious that this is causing the red lines to be pulled high, generating red pixels all the time.  Am I off-base here?  The beagleboard uses 10K-ohm resistors.  Is this enough to make a difference?  Is there something I'm missing, like the U-boot mux configuration or something?  What can I check from a software level to verify that all is correct?  According to the Beagle-board-xm manual, the board auto-detects which way things are hooked up, but maybe I haven't grabbed that code from the beagleboard board file?  Thanks,

Neil Johnson