My watch dog timer resets! Boy does it reset...I misread the datasheet and thought that the WDT was disabled after all resets (true only power-on reset). This, and the fact that a reset will change the timeout interval to the minimum caused the WDT to time out while cstartup was initializing stuff, before my code got a chance to run.

My carefully crafted test cases were examples of just how much time you could spend initializing variables before the minimum time WDT would time out. Also good examples of looking in the wrong direction.

I was briefly mislead thinking it had something to do with some WDT reset state interaction with the data pointer DPTR, and I tweeked the cstartup to use the alternate DPTR1 to initialize xdata variables...and that worked! I can't explain that one yet.

So let that be a lesson for you boys and girls, read those data sheets *very* carefully! I don't know how many times I read "The Watchdog timer reset does not disable the watchdog timer, but will restart it."