I'm not sure if this is a feature request or a bug, but I'm certain that it is a problem. Therefore, I'm creating a blank issue instead. Here is the problem I discovered:
Problem
The lat_unix_connect in LMbench will result in a connection refused error when running in Asterinas with LOG_LEVEL=error. However, it works well with LOG_LEVEL=debug. Then I found that the problem was that the server process failed to call the 'accept' syscall in time, causing the size of 'incoming_endpoints' in 'Backlog' to reach the 'backlog'.
Quick fix
After registering a timer callback function that calls schedule periodically with a 50ms frequency. The issue was resolved, and I received the results in about 1.8 microseconds.
I'm not sure if this is a feature request or a bug, but I'm certain that it is a problem. Therefore, I'm creating a blank issue instead. Here is the problem I discovered:
Problem
The
lat_unix_connect
in LMbench will result in a connection refused error when running in Asterinas withLOG_LEVEL=error
. However, it works well withLOG_LEVEL=debug
. Then I found that the problem was that the server process failed to call the 'accept' syscall in time, causing the size of 'incoming_endpoints' in 'Backlog' to reach the 'backlog'.Quick fix
After registering a timer callback function that calls
schedule
periodically with a 50ms frequency. The issue was resolved, and I received the results in about 1.8 microseconds.