Have you tried with another chip (preferrably from a different batch)? Does it still happen?
If only some chips do this, one possibility is that during production (of would-be QG8) some Flash fails the tests while the rest (within the QG4 range) passes. Freescale "baptises" the chip QG4, and you end up with a little "bonus" that you probably shouldn't trust (the extra memory, that is) for production but can be used for playing. They used to do that with HC11s. You ordered a ROM-less part to find either Buffalo or some other firmware in the on-chip ROM. [This actually saved me once in a production unit, when (for reasons too complicated to mention here) the part switched from expanded to single-chip mode and auto-executed the built-in Buffalo. I used Buffalo to set it back to using expanded mode.]
Then again, it could be a totally unrelated issue having to do with CW's debugging facilities of which I know nothing about as I don't use it (e.g., a "ghost" image that isn't really in the chip, only on the debugger's window).
Message Edited by tonyp on 2006-06-23 11:24 PM