Fixed an issue where, on single-core machines, the acceptThread could
get stuck due to the messageThread not yielding and hogging the fdsMtx lock. This caused long waiting times for new connections to be accept()ed.
Please register or sign in to comment