NOR flash erase failures - what do they look like?

If we lost power during a flash block erase on a traditional CFI NOR flash 
(e.g. S29AL032D) what would the resulting bit pattern likely look like? I 
know that the erase operation flips all bits to 1 so would it be possible to 
see an occasional 0 in a sea of 1s? Or would we see successive 0xFFs up to 
some point and then the previous content in the higher addresses?

Thanks,
Andrew

andrew queisser wrote:
> If we lost power during a flash block erase on a traditional CFI NOR flash 
> (e.g. S29AL032D) what would the resulting bit pattern likely look like? I 
> know that the erase operation flips all bits to 1 so would it be possible to 
> see an occasional 0 in a sea of 1s? Or would we see successive 0xFFs up to 
> some point and then the previous content in the higher addresses?

I don't know for certain but expect that the block is erased in parallel, 
so that you could see some, but not all bits changed from 0 to 1.  More 
worrisome is that some of the erased bits may be barely erased, susceptable 
to misreading later.

You could run an experiment of interrupting erasures and looking at memory.

-- 
Thad