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

[Peter Zijlstra]

On Mon, Feb 10, 2020 at 05:36:52AM -0800, Paul E. McKenney wrote:

> > Furthermore, using srcu would be detrimental, because of how it has
> > smp_mb() in the read side primitives.
> 
> Note that rcu_irq_enter() and rcu_irq_exit() also contain value-returning
> atomics, which imply full memory barriers.

There is a whole lot of perf that doesn't go through tracepoints. It
makes absolutely no sense to make all that more expensive just because
tracepoints are getting 'funny'.

> > The best we can do is move that rcu_irq_enter/exit_*() crud into the
> > perf tracepoint glue I suppose.
> 
> One approach would be to define a synchronize_preempt_disable() that
> waits only for pre-existing disabled-preemption regions (including
> of course diabled-irq and NMI-handler regions.  Something like Steve
> Rostedt's workqueue-baed schedule_on_each_cpu(ftrace_sync) implementation
> might work.
> 
> There are of course some plusses and minuses:
> 
> +	Works on preempt-disable regions in idle-loop code without
> 	the need to invoke rcu_idle_exit() and rcu_idle_enter()..
> 
> +	Straightforward implementation.
> 
> -	Does not work on preempt-disable regions on offline CPUs.
> 	(I have no idea if this really matters.)

I'd hope not ;-)

> -	Schedules on idle CPUs, so usage needs to be restricted to
> 	avoid messing up energy-efficient systems.  (It should be
> 	just fine to use this for tracing.)

Unless you're tracing energy usage -- weird some people actually do that
:-)