On 19 September 2026 17:30:26 BST, Linus Torvalds <[email protected]> wrote: >On Fri, 18 Sept 2026 at 17:01, Paul E. McKenney <[email protected]> >wrote: >> >> Its main benefit over RCU is that it allows fast reclaim of >> HP-protected pointers without needing to wait for a grace period. > >So last time I looked at this was some time ago, but the thing I was >missing then seems to still be missing: actual real numbers. > >And no, I most *definitely* do not mean the pointless "do this in a tight >loop". > >That's just garbage. Honestly, when I see those numbers it just makes >me go "I don't want to merge things that even mention this kind of >stupid load". It's worse than irrelevant - it's an actively >misleading. > >So I want to see a real load where this is actually visible not a >"loop a billion times on a hot-cache thing that doesn't do anything".
I wanna be a fanboy so bad here but I need to breathe and say something technical I think hazard pointers are good, what test do YOU suggest we do here? Once I get that PowerPC grant I'm happy to run it. God I wanted to be a fanboy and say omg Linus hi, but I'm sure I'll be flamed, sigh. Sigh. > > Linus > --- Thanks! "I'm not a very positive person" - Linus torvalds
