https://lwn.net/Articles/743714/
this is about a year old, but interesting in light of akaros's events. short version: it's a method to poll on linux AIO completions, using a ring buffer in shared memory. not sure if it ever got merged - i'm almost always a year behind on my LWN reading. one of the main things behind akaros's events and async syscalls was to avoid the proliferation of AIO/epoll/select mechanisms (and largely replace that with fast uthreads). the model they are using here is somewhat similar to requesting an event for an akaros syscall (though we allow someone to ask for the event *after* io_submit). the other thing they note is a bigger improvement from avoiding kernel crossings with the KPTI/KAISER mitigations. one of the comments talks about userspace's need to prioritize various events (e.g. FD completions for systemd). that seems like a nice job for a 2LS. barret -- You received this message because you are subscribed to the Google Groups "Akaros" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To post to this group, send email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/akaros/20190104115306.4389ee87%40brho.roam.corp.google.com. For more options, visit https://groups.google.com/d/optout.
