Linux 5.17 To Boast Latency Optimization For AF_UNIX Sockets

Net-next has been queuing a number of enticing performance optimizations ahead of the Linux 5.17 merge window kicking off around the start of the new year. Covered already was a big TCP optimization and a big improvement for csum_partial() that is used in the network code for checksum computation. The latest optimization is improving the AF_UNIX code path for those using AF_UNIX sockets for local inter-process communication.

A new patch series was queued up on Friday in net-next for improving the AF_UNIX code. That patch series by Kuniyuki Iwashima of Amazon Japan is ultimately about replacing AF_UNIX sockets' single big lock with per-hash locks. The series replaces the AF_UNIX big lock and also as part of the series has a speed-up to the autobind behavior.

The patch series is enticing for heavy AF_UNIX usage. Replacing the big lock has a nice latency improvement in testing conducted by Amazon AWS:

Relaxing a race within unix_autobind() provides for another nice latency improvement in cases of autobind'ing many sockets in parallel by searching for a name from a random number.

Linux 5.17 material continues piling up to make for an exciting next kernel cycle in 2022.