Re: [PATCH 2/2] tracing: Use proper do_arch_spin_lock() API

[Peter Zijlstra]

I'm with Steve on not adding wrappers for this; people that use
arch_spinlock_* get to keep the pieces :-)

On Wed, Sep 21, 2022 at 06:17:21PM -0400, Steven Rostedt wrote:

> > @@ -1819,7 +1819,7 @@ update_max_tr(struct trace_array *tr, struct task_struct *tsk, int cpu,
> >  		return;
> >  	}
> >  
> > -	arch_spin_lock(&tr->max_lock);
> > +	do_arch_spin_lock(&tr->max_lock);
> 
> Nothing here is needed, as interrupts had better be disabled when this
> function is called. And there's already a:
> 
> 	WARN_ON_ONCE(!irqs_disabled());

You can write that as lockdep_assert_irqs_disabled();