On 2026-09-19 13:00, Linus Torvalds wrote: [...]
Now, obviously, the thing that would impress me is something like the dcache. If *that* can be converted, and it shows real improvements on some real benchmark, then I'm sold.
[ Side-discussion, not related to hazard pointers. ] Talking about dcache improvements: I've been working on scalability of rename, directory move, and directory listing, basically killing dcache global locks, seqlocks and rwlocks which significantly limit dcache update and read-side scalability. Those improvements are based on RCU, not hazard pointers, and on a new "RCU pseudo-transaction" concept I've invented earlier this year. I'm preparing a paper and I plan to present it at LPC in Prague. My benchmark results are on a userspace port of the Linux kernel dentry cache at this point, so it's not ready for anything close to upstream at this stage. Cheers, Mathieu -- Mathieu Desnoyers EfficiOS Inc. https://www.efficios.com
