Re: [PATCH] printk: Make console tracepoint safe in NMI() context

[Steven Rostedt]

On Fri, 15 Jul 2022 14:01:52 +0200
Petr Mladek <pmladek@xxxxxxxx> wrote:

> +	/*
> +	 * trace_console_rcuidle() is not working in NMI. printk()
> +	 * is used more often in NMI than in rcuidle context.
> +	 * Choose the less evil solution here.
> +	 *
> +	 * raw_smp_processor_id() is reliable in rcuidle context.
> +	 */
> +	TP_CONDITION(!rcu_is_idle_cpu(raw_smp_processor_id())),
> +

As Marco mentioned in the other thread, would a check for
'rcu_is_watching()' be better?

-- Steve


>  	TP_STRUCT__entry(
>  		__dynamic_array(char, msg, len + 1)
>  	),