As it turns out so often, the problem was nothing to do with my original theory about the reset line.

What _was_ happening that we were not handling the SPIF=0 clear for all characters received by the SPI interrupt handler only the ones we are acting on. At startup there were one or more garbage characters sent by the QSPI module on the 5272 as it boots up (I guess). Since these were not being handled, I was not clearing the SPI interrupt and the program essentially was just sitting in the SPI interrupt handler.

The other thing we were not checking for was overrun of the SPI buffer, we have also fixed that. Obvious things but....


Thank you all for your help

John Robbins