Closed xThaid closed 6 months ago
If I correctly understand the bug was that in case when start_refill
was not ready, then MemRead
transaction wasn't started and this caused that issue_req
could be called second time, what caused that data from first call were overwritten.
If I correctly understand the bug was that in case when
start_refill
was not ready, thenMemRead
transaction wasn't started and this caused thatissue_req
could be called second time, what caused that data from first call were overwritten.
Exactly.
The instruction cache code didn't correctly handle a case when the
start_refill
method of the cache refiller was not active. The tests didn't test it at all, asstart_refill
was always ready there.