Proposed testcase: Start flashing and interrupt that in the middle

[fhaverkamp]

Hi Kenneth,

I have seen cases where we accidentally started flashing the wrong bitstream. We noticed and pressed
Cntl+C. The bad thing was, that trying to flash the right data did not work and the erase timed out as
somehow noticed in other cases. Just as if the flash block is protected and not released by the programming process. We used the USB bitstream load tool and a fill power cycle to recover the card.

I would assume that the flash update should recover on the 2nd attempt unless the card looses the bitstream. I do not like to repro this myself, unless it happens again accidently. But maybe you find some
time to explore how it behaves for your cards.

I furthermore assume that your flashing tool cannot do much here and the problem might lie in the PSL part responsible for flashing. If you happen to repro this nicely, you might at least complain that PSL folks can consider fixing that for future cards, or once updating PSLs for other reasons. Loosing cards due to this in production environments is really something ugly I suppose.

Regards

Frank

[kenhill]

@fhaverkamp master currently traps and resets the CAPI adapter to the factory partition. There was a bug in the PSL logic that was preventing successful perst to user/factory. A fix is being pushed out into the DCP files soon