Details
-
Improvement
-
Status: Closed
-
Major
-
Resolution: Fixed
-
None
-
None
Description
There is significant complexity in the proactor epoll locking - to the extent that it is (or at least I find it) hard to reason about where locks are taken and release.
Some of this complexity is due to the locks not being strictly hierarchical and some of them are taken in one place and then released in another.
There may be a performance justification for this complexity, but I would like to simplify the locking and then improve it if it is found to be a real performance issue.