[PATCH v6 1/4] Add LLL_MUTEX_READ_LOCK [BZ #28537]

[Noah Goldstein]

On Sun, Sep 11, 2022 at 4:19 PM Sunil Pandey <skpgkp2@gmail.com> wrote:
>
> On Wed, Nov 17, 2021 at 5:17 PM Arjan van de Ven via Libc-alpha
> <libc-alpha@sourceware.org> wrote:
> >
> > On 11/17/2021 4:31 PM, H.J. Lu wrote:
> > >> Yes, but the loop will be able to run `max_cnt` iterations much faster now.
> > >> Just wondering if the value needs to be re-tuned. Not that is necessarily needs
> > >> to be.
> > > Maybe if we can find some data to show for.
> > >
> >
> > wondering when this was last tuned.. I assume todays CPUs and CPUs from, say, 5 or 10 years ago
> > have an order of magnitude different performance in this regard....
> > if there wasn't a need to retune during that, maybe this value is so robust that it doesn't need
> > retuning.
> >
> > or maybe it's time to retune this in general sometime soon after this patch goes in ;)
>
> I would like to backport this patch to release branch 2.33 and 2.34

Fine by  me.
>
> Any comments/suggestions or objections on this.
>
> commit d672a98a1af106bd68deb15576710cd61363f7a6
> Author: H.J. Lu <hjl.tools@gmail.com>
> Date:   Tue Nov 2 18:33:07 2021 -0700
>
>     Add LLL_MUTEX_READ_LOCK [BZ #28537]
>
>     CAS instruction is expensive.  From the x86 CPU's point of view, getting
>     a cache line for writing is more expensive than reading.  See Appendix
>     A.2 Spinlock in:
>
>     https://www.intel.com/content/dam/www/public/us/en/documents/white-papers/xeon-lock-scaling-analysis-paper.pdf