EIM injects error to the particular RAM array bus, so it is being detected by the side affected by this error.
If it is write (thus transfer from Core0 to PRAMC_0 with error on the way), it’ll be detected by PRAMC_0.
If it is read (thus transfer from PRAMC_0 to Core0 with error on the way), it’ll be detected by Core0.
If you see more errors in more ERM channels at once, if could be possibly caused by enabled cache memory.