LWN: Comments on "Redesigned workqueues for io_uring" https://lwn.net/Articles/803070/ This is a special feed containing comments posted to the individual LWN article titled "Redesigned workqueues for io_uring". en-us Sat, 25 Oct 2025 23:35:42 +0000 Sat, 25 Oct 2025 23:35:42 +0000 https://www.rssboard.org/rss-specification lwn@lwn.net Redesigned workqueues for io_uring https://lwn.net/Articles/803323/ https://lwn.net/Articles/803323/ ncultra <div class="FormattedComment"> elegant :)<br> </div> Mon, 28 Oct 2019 15:31:50 +0000 Redesigned workqueues for io_uring https://lwn.net/Articles/803229/ https://lwn.net/Articles/803229/ axboe <div class="FormattedComment"> You can, the hashing is only for writers.<br> </div> Sat, 26 Oct 2019 14:55:34 +0000 Redesigned workqueues for io_uring https://lwn.net/Articles/803222/ https://lwn.net/Articles/803222/ MattBBaker <div class="FormattedComment"> It looks like that is just for buffered ops. Which seems reasonable, most applications that would want such an interface are likely playing with the DIO power tools anyways.<br> </div> Sat, 26 Oct 2019 12:26:54 +0000 Redesigned workqueues for io_uring https://lwn.net/Articles/803195/ https://lwn.net/Articles/803195/ pbonzini <div class="FormattedComment"> Most of the work is done asynchronously, but the remaining part would encounter contention.<br> </div> Sat, 26 Oct 2019 08:45:26 +0000 Redesigned workqueues for io_uring https://lwn.net/Articles/803177/ https://lwn.net/Articles/803177/ roc <div class="FormattedComment"> <font class="QuotedText">&gt; If an application submits multiple buffered I/O requests for a single file, they should not be run concurrently or they are likely to just block each other via lock contention.</font><br> <p> Does this mean you can't issue multiple reads from the same file and have them run concurrently? That would be sad.<br> </div> Fri, 25 Oct 2019 19:54:44 +0000