Brian,
Thanks for your detailed answers. I have a few more questions:
What version of LMFlash are you using? If you are not using the latest version, get it from here and try it.
Can you share how often you seeing this problem? If this is a common event, we can try to reproduce it at TI.
Does this still happen when you use the GUI version of LMFlash? In some cases LMFlash has better error reporting in the GUI version.
When you say that the you "run the board again" does this mean that you power it off and on?
In answer to your questions:
1. There is not a specified settling time for the ICDI. Any code running on the ICDI is likely limited by the enumeration process of the host PC.
2. Some error conditions in LMFlash will return more descriptive error messages about the error source that can then be logged. If LMFlash returns a 1 though, it does hinder the investigation.
3. LMFlash provides other exit codes for some other functions, but I looked at the hardware reset function, and it only returns a 1.
4. I know of no reason why a programmed part would take longer to come up. The ICDI does not communicate with the target part until prompted by the host.