Since 2019 Linux has io_uring, and the last few releases expanded it. io_uring is a pretty exciting mechanism that is rapidly becoming the highest performance m

Since 2019 Linux has io_uring, and the last few releases expanded it. io_uring is a pretty exciting mechanism that is rapidly becoming the highest performance mechanism for doing syscalls in Linux. This is very important because spectre/meltdown mitigations made syscalls much slower; Linux really need truly async syscalls to regain some losses here. This also mirrors Google's new OS Fuschia, in which all syscalls are async. Here are some articles Ringing in a new asynchronous I/O API (January 2019) https://lwn.net/Articles/776703/ https://lwn.net/Articles/776703/ The rapid growth of io_uring (January 2020) https://lwn.net/Articles/810414/ https://lwn.net/Articles/810414/ BPF meets io_uring (March 2021) https://lwn.net/Articles/847951/ https://lwn.net/Articles/847951/ Auditing io_uring (July 2021) https://lwn.net/Articles/858023/ https://lwn.net/Articles/858023/ Descriptorless files for io_uring (July 2021) https://lwn.net/Articles/863071/ https://lwn.net/Articles/863071/