Re: [RFC 0/3] Revert SRCU from tracepoint infrastructure

[Paul E. McKenney]

On Mon, Feb 10, 2020 at 12:05:52PM -0500, Steven Rostedt wrote:
> On Mon, 10 Feb 2020 10:46:16 +0100
> Peter Zijlstra <peterz@xxxxxxxxxxxxx> wrote:
> 
> > Furthermore, using srcu would be detrimental, because of how it has
> > smp_mb() in the read side primitives.
> 
> I didn't realize that there was a full memory barrier in the srcu read
> side. Seems to me that itself is rational for reverting it. And also a
> big NAK for any suggestion to have any of the function tracing to use
> it as well (which comes up here and there).

Yeah, that was added some years back when people were complaining about
three synchronize_sched()'s worth of latency for synchronize_srcu().

I did prepare a patch about a year ago that would allow an srcu_struct
to be set up to not need the read-side smp_mb() calls, but this means
longer-latency grace periods (though nowhere near as long as those of
synchronize_rcu_tasks()) and also that the "fast SRCU" readers cannot
be used from idle or offline CPUs.

But an easy change if it is useful.

							Thanx, Paul