You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The old hashmap allowed concurrent reads.
The new hashmap code didn't, and led to crashes when GOMAXPROCS > 1.
From khr:
-----------
So I think we've found the problem. It is triggered by two parallel reads to a map. My
hashmap implementation was assuming that wouldn't happen. Internally, the hash table is
grown lazily and I was doing a little bit of growing work during each read. That
growing work can't happen in parallel and causes corruption.
The quick fix is to disable growing work on reads. I'll try to get that in today. The
longer fix would be to have some internal synchronization for the growing work.
------------
Several other bugs may or may not be this issue.
The text was updated successfully, but these errors were encountered:
The text was updated successfully, but these errors were encountered: