Re: [PATCH v4 4/6] tracing: Guard __DECLARE_TRACE() use of __DO_TRACE_CALL() with SRCU-fast

[Paul E. McKenney]

On Wed, Jul 23, 2025 at 06:29:30PM -0400, Steven Rostedt wrote:
> On Wed, 23 Jul 2025 15:17:52 -0700
> "Paul E. McKenney" <paulmck@xxxxxxxxxx> wrote:
> 
> > I believe that Steve provided me with the essentials for perf and ftrace,
> > but please check:  f808f53d4e4f ("squash! tracing: Guard __DECLARE_TRACE()
> > use of __DO_TRACE_CALL() with SRCU-fast").
> 
> Note, there's nothing in the ftrace side that requires preemption disabled,
> but it assumes that it is, and adjusts the preempt_count that is recorded
> in the trace event to accommodate it.

Ah, thank you for the clarification.  I agree with your approach as being
a more localized change, with less chance of some forgotten invariant
biting us.  ;-)

							Thanx, Paul