Re: [PATCH 2/3] locking: Clarify requirements for smp_mb__after_spinlock()

From: Andrea Parri
Date: Thu Jun 28 2018 - 13:31:00 EST


On Thu, Jun 28, 2018 at 05:05:50PM +0200, Peter Zijlstra wrote:
> On Thu, Jun 28, 2018 at 12:41:19PM +0200, Andrea Parri wrote:
> > - * This barrier must provide two things:
> > - *
> > - * - it must guarantee a STORE before the spin_lock() is ordered against a
> > - * LOAD after it, see the comments at its two usage sites.
> > - *
> > - * - it must ensure the critical section is RCsc.
> > - *
> > - * The latter is important for cases where we observe values written by other
> > - * CPUs in spin-loops, without barriers, while being subject to scheduling.
> > - *
> > - * CPU0 CPU1 CPU2
> > - *
> > - * for (;;) {
> > - * if (READ_ONCE(X))
> > - * break;
> > - * }
> > - * X=1
> > - * <sched-out>
> > - * <sched-in>
> > - * r = X;
> > - *
> > - * without transitivity it could be that CPU1 observes X!=0 breaks the loop,
> > - * we get migrated and CPU2 sees X==0.
>
> Please don't remove that; that explains _why_ we need a full memory
> barrier here.

Peter:

Both you and Boqun stated that the above snippet is "bad":

http://lkml.kernel.org/r/20180312085646.GE4064@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

and I do agree with your assessment! ;-)

I've no objection to keep that comment (together with the
"clarification" suggested in this patch) _once_ replaced
that snippet with something else (say, with the snippet
Boqun suggested in:

http://lkml.kernel.org/r/20180312085600.aczjkpn73axzs2sb@tardis ):

is this what you mean?

Andrea