Sam,

Have you tried using bootme & instead of load?
I'd be very interested to see if you still get the same results that way.
Thanks

Sam

The reset button on the OMAP35x kit is tied to the processor's SYS_nRESPWRON, so it performs a power on reset. What it does not do, is reset the PMIC. Therefore, when you use the reset button, only the processor gets reset and the PMIC does not. That means all voltages remain valid/stable during a reset and the device memory could be held. Instead of actually removing the power cable each time, can you just use the Power On/Off button, S3? It should provide the same functionality.
Another possibility is that since the PMIC does not get reset, there are still valid voltages on the OMAP processor; I know this processor has a complex Boot ROM that functions differently if powering from no voltage, backup voltage, software reset, hardware reset, etc. It's possible that your software needs to program the Boot ROM such that it does a full reset when SYS_nRESPWRON is pressed.

Nate

Nathan-thanks for the tip, I'll use the power button from now on.
Richard-I haven't tried this yet, although I just got KITL up and going today, so I'll keep you updated if I notice/don't notice this.

The thing I don't understand is why LoLo wouldn't overwrite the prior image in SDRAM upon the reload after the reset. It's like the load command checked to see if the OS image was still in SDRAM, and then put the new one in a different location as to not overwrite it. But then "exec" stuck with booting the original one.