I did some tests in EMR polluted environment and the SAME70Q21 (384kB SRAM) crashed badly. I wasn't able to bring the chip back to life by only attaching the Atmel ICE debugger and running the Atmel Studio debugger without power on rebooting the chip.
Without cold reboot I got the following error message in AS:
"Unexpected Chip Identifier 0xa10d0e00 (expected 0xa1020e00)"
After power on reset this error disappeared which means that the correct chip ID was read again.
I was able to reproduce this easily multiple times and got always the same error message and chip ID values.
Seems like byte 4 == SRAMSIZ in CHIPID_CIDR fell over due to the EMR shock from 0x2 == 384kB to 0xd == 256kB and was therefore not recognized by AS.
I wonder how this is possible because according to the manual this ID is kind of hard-wired in the chip. And according to the data sheet this is not a valid E70 chip ID / chip type.
Any idea ? Is this a bug in Atmel-ICE or AS7 ?
P.S. what really hurts is that the chip did not restart the application although watchdog, brownout reset, clock and supply voltage failure detection are enabled. I tested each of these supervisor mechanisms with success. Anything I can do further ?