Details:

Potential race condition in capture registers resulting in events getting lost while other events in the same register are being cleared by writing to the register. Following registers are impacted by this issue:

Workaround(s):

Any of the following Workarounds can be used:

Sequentially trigger the mem-init and clear the status before triggering the new mem-init. This is needed if both the status are in the same register.

(OR)

If parallel triggers are must then poll for the all status-bits that got triggered to be 1'b1 and then go and clear the DONE status register

(OR)

Check the MEM_INIT_STATUS register after starting the mem-init and wait the status to go -low by checking it in regular interval and finally clear the DONE status register when the status goes low