ChanServ changed the topic of #io_uring to: io_uring development discussion | Logs: https://oftc.irclog.whitequark.org/io_uring
raphaelsc has quit [Remote host closed the connection]
raphaelsc has joined #io_uring
rnsanchez has joined #io_uring
<rnsanchez> I've found myself in a loop upon calling io_uring_wait_cqe(). it's a new prototype so I was expecting to get into a sleep state (nothing issued onto the ring). after a much-longer debugging session, I couldn't remember why I had a io_uring_buf_ring_cq_advance() call after setting up buffers in a loop with io_uring_buf_ring_add(). is that expected? I understand I've probably misused liburing's API. the returned CQEs were always
<rnsanchez> completely zeroed out
<rnsanchez> as soon as I removed the call to io_uring_buf_ring_cq_advance() after that loop setup, things worked as expected: process sleeping upon calling io_uring_wait_cqe() and CQEs properly filled out
<rnsanchez> apparently, I should have used io_uring_buf_ring_advance() instead of io_uring_buf_ring_cq_advance() (per io_uring_and_networking_in_2023 PDF)
raphaelsc has quit [Remote host closed the connection]
raphaelsc has joined #io_uring